{"doc_key": "ai-train-1", "ner": [[4, 10, "product"], [17, 19, "field"], [22, 24, "task"], [27, 29, "task"], [34, 37, "task"], [41, 42, "field"], [43, 44, "researcher"], [46, 47, "researcher"], [49, 50, "researcher"], [52, 53, "researcher"], [55, 56, "researcher"], [58, 59, "researcher"], [61, 62, "researcher"], [64, 65, "researcher"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13], "relations": [[4, 10, 17, 19, "part-of", "", false, false], [4, 10, 17, 19, "usage", "", false, false], [4, 10, 22, 24, "part-of", "", false, false], [4, 10, 22, 24, "usage", "", false, false], [4, 10, 27, 29, "part-of", "", false, false], [4, 10, 27, 29, "usage", "", false, false], [4, 10, 41, 42, "part-of", "", false, false], [4, 10, 41, 42, "usage", "", false, false], [34, 37, 27, 29, "part-of", "", false, false], [34, 37, 27, 29, "type-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "sentence": ["Les", "approches", "populaires", "du", "syst\u00e8me", "de", "recommandation", "bas\u00e9", "sur", "l'", "opinion", "utilisent", "diverses", "techniques", ",", "notamment", "l'", "exploration", "de", "texte", ",", "la", "recherche", "d'", "informations", ",", "l'", "analyse", "des", "sentiments", "(", "voir", "\u00e9galement", "l'", "analyse", "multimodale", "des", "sentiments", ")", "et", "l'", "apprentissage", "profond", "X.Y.", "Feng", ",", "H.", "Zhang", ",", "Y.J.", "Ren", ",", "P.H.", "Shang", ",", "Y.", "Zhu", ",", "Y.C.", "Liang", ",", "R.C.", "Guan", ",", "D.", "Xu", ",", "(", "2019", ")", ",", ",", "21", "(", "5", ")", ":", "e12957", "."], "sentence-detokenized": "Les approches populaires du syst\u00e8me de recommandation bas\u00e9 sur l'opinion utilisent diverses techniques, notamment l'exploration de texte, la recherche d'informations, l'analyse des sentiments (voir \u00e9galement l'analyse multimodale des sentiments) et l'apprentissage profond X.Y. Feng, H. Zhang, Y.J. Ren, P.H. Shang, Y. Zhu, Y.C. Liang, R.C. Guan, D. Xu, (2019),, 21 (5) : e12957.", "token2charspan": [[0, 3], [4, 13], [14, 24], [25, 27], [28, 35], [36, 38], [39, 53], [54, 58], [59, 62], [63, 65], [65, 72], [73, 82], [83, 91], [92, 102], [102, 103], [104, 113], [114, 116], [116, 127], [128, 130], [131, 136], [136, 137], [138, 140], [141, 150], [151, 153], [153, 165], [165, 166], [167, 169], [169, 176], [177, 180], [181, 191], [192, 193], [193, 197], [198, 207], [208, 210], [210, 217], [218, 229], [230, 233], [234, 244], [244, 245], [246, 248], [249, 251], [251, 264], [265, 272], [273, 277], [278, 282], [282, 283], [284, 286], [287, 292], [292, 293], [294, 298], [299, 302], [302, 303], [304, 308], [309, 314], [314, 315], [316, 318], [319, 322], [322, 323], [324, 328], [329, 334], [334, 335], [336, 340], [341, 345], [345, 346], [347, 349], [350, 352], [352, 353], [354, 355], [355, 359], [359, 360], [360, 361], [361, 362], [363, 365], [366, 367], [367, 368], [368, 369], [370, 371], [372, 378], [378, 379]]}
{"doc_key": "ai-train-2", "ner": [[9, 9, "university"], [15, 16, "researcher"], [18, 19, "researcher"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[15, 16, 9, 9, "physical", "", false, false], [15, 16, 9, 9, "role", "", false, false], [18, 19, 9, 9, "physical", "", false, false], [18, 19, 9, 9, "role", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Les", "d\u00e9fenseurs", "des", "repr\u00e9sentations", "proc\u00e9durales", "\u00e9taient", "principalement", "centr\u00e9s", "au", "MIT", ",", "sous", "la", "direction", "de", "Marvin", "Minsky", "et", "Seymour", "Papert", "."], "sentence-detokenized": "Les d\u00e9fenseurs des repr\u00e9sentations proc\u00e9durales \u00e9taient principalement centr\u00e9s au MIT, sous la direction de Marvin Minsky et Seymour Papert.", "token2charspan": [[0, 3], [4, 14], [15, 18], [19, 34], [35, 47], [48, 55], [56, 70], [71, 78], [79, 81], [82, 85], [85, 86], [87, 91], [92, 94], [95, 104], [105, 107], [108, 114], [115, 121], [122, 124], [125, 132], [133, 139], [139, 140]]}
{"doc_key": "ai-train-3", "ner": [[12, 12, "programlang"]], "ner_mapping_to_source": [0], "relations": [], "relations_mapping_to_source": [], "sentence": ["L'", "interface", "standard", "et", "l'", "interface", "de", "la", "calculatrice", "sont", "\u00e9crites", "en", "Java", "."], "sentence-detokenized": "L'interface standard et l'interface de la calculatrice sont \u00e9crites en Java.", "token2charspan": [[0, 2], [2, 11], [12, 20], [21, 23], [24, 26], [26, 35], [36, 38], [39, 41], [42, 54], [55, 59], [60, 67], [68, 70], [71, 75], [75, 76]]}
{"doc_key": "ai-train-4", "ner": [[0, 0, "product"], [32, 32, "product"]], "ner_mapping_to_source": [0, 1], "relations": [[0, 0, 32, 32, "related-to", "compatible_with", false, false]], "relations_mapping_to_source": [0], "sentence": ["Octave", "permet", "de", "r\u00e9soudre", "num\u00e9riquement", "des", "probl\u00e8mes", "lin\u00e9aires", "et", "non", "lin\u00e9aires", ",", "et", "d'", "effectuer", "d'", "autres", "exp\u00e9riences", "num\u00e9riques", "\u00e0", "l'", "aide", "d'", "un", "logiciel", "qui", "est", "en", "grande", "partie", "compatible", "avec", "MATLAB", "."], "sentence-detokenized": "Octave permet de r\u00e9soudre num\u00e9riquement des probl\u00e8mes lin\u00e9aires et non lin\u00e9aires, et d'effectuer d'autres exp\u00e9riences num\u00e9riques \u00e0 l'aide d'un logiciel qui est en grande partie compatible avec MATLAB.", "token2charspan": [[0, 6], [7, 13], [14, 16], [17, 25], [26, 39], [40, 43], [44, 53], [54, 63], [64, 66], [67, 70], [71, 80], [80, 81], [82, 84], [85, 87], [87, 96], [97, 99], [99, 105], [106, 117], [118, 128], [129, 130], [131, 133], [133, 137], [138, 140], [140, 142], [143, 151], [152, 155], [156, 159], [160, 162], [163, 169], [170, 176], [177, 187], [188, 192], [193, 199], [199, 200]]}
{"doc_key": "ai-train-5", "ner": [[4, 6, "algorithm"], [10, 12, "misc"], [14, 15, "researcher"], [22, 24, "university"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[4, 6, 14, 15, "origin", "", false, false], [10, 12, 14, 15, "origin", "", false, false], [14, 15, 22, 24, "physical", "", false, false], [14, 15, 22, 24, "role", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Des", "variantes", "de", "l'", "algorithme", "de", "r\u00e9tropropagation", "ainsi", "que", "les", "m\u00e9thodes", "non", "supervis\u00e9es", "de", "Geoff", "Hinton", "et", "de", "ses", "coll\u00e8gues", "de", "l'", "universit\u00e9", "de", "Toronto", "peuvent", "\u00eatre", "utilis\u00e9es", "pour", "former", "des", "architectures", "neuronales", "profondes", "et", "hautement", "non", "lin\u00e9aires", ",", "{{cite", "journal"], "sentence-detokenized": "Des variantes de l'algorithme de r\u00e9tropropagation ainsi que les m\u00e9thodes non supervis\u00e9es de Geoff Hinton et de ses coll\u00e8gues de l'universit\u00e9 de Toronto peuvent \u00eatre utilis\u00e9es pour former des architectures neuronales profondes et hautement non lin\u00e9aires, {{cite journal", "token2charspan": [[0, 3], [4, 13], [14, 16], [17, 19], [19, 29], [30, 32], [33, 49], [50, 55], [56, 59], [60, 63], [64, 72], [73, 76], [77, 88], [89, 91], [92, 97], [98, 104], [105, 107], [108, 110], [111, 114], [115, 124], [125, 127], [128, 130], [130, 140], [141, 143], [144, 151], [152, 159], [160, 164], [165, 174], [175, 179], [180, 186], [187, 190], [191, 204], [205, 215], [216, 225], [226, 228], [229, 238], [239, 242], [243, 252], [252, 253], [254, 260], [261, 268]]}
{"doc_key": "ai-train-6", "ner": [[8, 8, "metrics"]], "ner_mapping_to_source": [0], "relations": [], "relations_mapping_to_source": [], "sentence": ["ou", "de", "mani\u00e8re", "\u00e9quivalente", "en", "utilisant", "la", "notation", "DCG", ":"], "sentence-detokenized": "ou de mani\u00e8re \u00e9quivalente en utilisant la notation DCG :", "token2charspan": [[0, 2], [3, 5], [6, 13], [14, 25], [26, 28], [29, 38], [39, 41], [42, 50], [51, 54], [55, 56]]}
{"doc_key": "ai-train-7", "ner": [[0, 2, "algorithm"], [6, 8, "algorithm"], [16, 19, "algorithm"], [23, 27, "algorithm"], [31, 31, "algorithm"], [33, 35, "algorithm"], [52, 53, "misc"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "relations": [[0, 2, 6, 8, "type-of", "", false, false], [0, 2, 16, 19, "usage", "part-of?", true, false], [16, 19, 23, 27, "compare", "", false, false], [31, 31, 23, 27, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Les", "cartes", "auto-organisatrices", "diff\u00e8rent", "des", "autres", "r\u00e9seaux", "neuronaux", "artificiels", "dans", "la", "mesure", "o\u00f9", "elles", "appliquent", "un", "apprentissage", "comp\u00e9titif", ",", "par", "opposition", "\u00e0", "l'", "apprentissage", "par", "correction", "d'", "erreurs", "(", "comme", "la", "r\u00e9tropropagation", "avec", "descente", "de", "gradient", ")", ",", "et", "dans", "la", "mesure", "o\u00f9", "elles", "utilisent", "une", "fonction", "de", "voisinage", "pour", "pr\u00e9server", "les", "propri\u00e9t\u00e9s", "topologiques", "de", "l'", "espace", "d'", "entr\u00e9e", "."], "sentence-detokenized": "Les cartes auto-organisatrices diff\u00e8rent des autres r\u00e9seaux neuronaux artificiels dans la mesure o\u00f9 elles appliquent un apprentissage comp\u00e9titif, par opposition \u00e0 l'apprentissage par correction d'erreurs (comme la r\u00e9tropropagation avec descente de gradient), et dans la mesure o\u00f9 elles utilisent une fonction de voisinage pour pr\u00e9server les propri\u00e9t\u00e9s topologiques de l'espace d'entr\u00e9e.", "token2charspan": [[0, 3], [4, 10], [11, 30], [31, 40], [41, 44], [45, 51], [52, 59], [60, 69], [70, 81], [82, 86], [87, 89], [90, 96], [97, 99], [100, 105], [106, 116], [117, 119], [120, 133], [134, 144], [144, 145], [146, 149], [150, 160], [161, 162], [163, 165], [165, 178], [179, 182], [183, 193], [194, 196], [196, 203], [204, 205], [205, 210], [211, 213], [214, 230], [231, 235], [236, 244], [245, 247], [248, 256], [256, 257], [257, 258], [259, 261], [262, 266], [267, 269], [270, 276], [277, 279], [280, 285], [286, 295], [296, 299], [300, 308], [309, 311], [312, 321], [322, 326], [327, 336], [337, 340], [341, 351], [352, 364], [365, 367], [368, 370], [370, 376], [377, 379], [379, 385], [385, 386]]}
{"doc_key": "ai-train-8", "ner": [[12, 14, "organisation"], [30, 31, "misc"], [39, 43, "metrics"]], "ner_mapping_to_source": [0, 1, 2], "relations": [], "relations_mapping_to_source": [], "sentence": ["Depuis", "le", "d\u00e9but", "des", "ann\u00e9es", "1990", ",", "plusieurs", "organismes", ",", "dont", "l'", "Audio", "Engineering", "Society", ",", "recommandent", "que", "les", "mesures", "de", "la", "gamme", "dynamique", "soient", "effectu\u00e9es", "en", "pr\u00e9sence", "d'", "un", "signal", "audio", ",", "qui", "est", "ensuite", "filtr\u00e9", "dans", "la", "mesure", "du", "plancher", "de", "bruit", "utilis\u00e9e", "pour", "d\u00e9terminer", "la", "gamme", "dynamique", ".", "Cela", "permet", "d'", "\u00e9viter", "les", "mesures", "douteuses", "bas\u00e9es", "sur", "l'", "utilisation", "de", "supports", "vierges", "ou", "de", "circuits", "d'", "inhibition", "."], "sentence-detokenized": "Depuis le d\u00e9but des ann\u00e9es 1990, plusieurs organismes, dont l'Audio Engineering Society, recommandent que les mesures de la gamme dynamique soient effectu\u00e9es en pr\u00e9sence d'un signal audio, qui est ensuite filtr\u00e9 dans la mesure du plancher de bruit utilis\u00e9e pour d\u00e9terminer la gamme dynamique. Cela permet d'\u00e9viter les mesures douteuses bas\u00e9es sur l'utilisation de supports vierges ou de circuits d'inhibition.", "token2charspan": [[0, 6], [7, 9], [10, 15], [16, 19], [20, 26], [27, 31], [31, 32], [33, 42], [43, 53], [53, 54], [55, 59], [60, 62], [62, 67], [68, 79], [80, 87], [87, 88], [89, 101], [102, 105], [106, 109], [110, 117], [118, 120], [121, 123], [124, 129], [130, 139], [140, 146], [147, 157], [158, 160], [161, 169], [170, 172], [172, 174], [175, 181], [182, 187], [187, 188], [189, 192], [193, 196], [197, 204], [205, 211], [212, 216], [217, 219], [220, 226], [227, 229], [230, 238], [239, 241], [242, 247], [248, 256], [257, 261], [262, 272], [273, 275], [276, 281], [282, 291], [291, 292], [293, 297], [298, 304], [305, 307], [307, 313], [314, 317], [318, 325], [326, 335], [336, 342], [343, 346], [347, 349], [349, 360], [361, 363], [364, 372], [373, 380], [381, 383], [384, 386], [387, 395], [396, 398], [398, 408], [408, 409]]}
{"doc_key": "ai-train-9", "ner": [[6, 7, "misc"], [21, 23, "task"], [25, 29, "task"], [31, 32, "task"], [34, 35, "task"], [39, 40, "task"], [37, 46, "task"], [48, 52, "field"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "relations": [[6, 7, 21, 23, "part-of", "concept_used_in", true, false], [6, 7, 25, 29, "part-of", "concept_used_in", false, false], [6, 7, 31, 32, "part-of", "concept_used_in", false, false], [6, 7, 34, 35, "part-of", "concept_used_in", false, false], [6, 7, 39, 40, "part-of", "concept_used_in", false, false], [6, 7, 37, 46, "part-of", "concept_used_in", false, false], [6, 7, 48, 52, "part-of", "concept_used_in", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "sentence": ["La", "technique", "utilis\u00e9e", "pour", "cr\u00e9er", "des", "visages", "propres", "et", "les", "utiliser", "pour", "la", "reconnaissance", "est", "\u00e9galement", "utilis\u00e9e", "en", "dehors", "de", "la", "reconnaissance", "des", "visages", ":", "reconnaissance", "de", "l'", "\u00e9criture", "manuscrite", ",", "lecture", "labiale", ",", "reconnaissance", "vocale", ",", "interpr\u00e9tation", "du", "langage", "des", "signes", "/", "des", "gestes", "des", "mains", "et", "analyse", "de", "l'", "imagerie", "m\u00e9dicale", "."], "sentence-detokenized": "La technique utilis\u00e9e pour cr\u00e9er des visages propres et les utiliser pour la reconnaissance est \u00e9galement utilis\u00e9e en dehors de la reconnaissance des visages : reconnaissance de l'\u00e9criture manuscrite, lecture labiale, reconnaissance vocale, interpr\u00e9tation du langage des signes/des gestes des mains et analyse de l'imagerie m\u00e9dicale.", "token2charspan": [[0, 2], [3, 12], [13, 21], [22, 26], [27, 32], [33, 36], [37, 44], [45, 52], [53, 55], [56, 59], [60, 68], [69, 73], [74, 76], [77, 91], [92, 95], [96, 105], [106, 114], [115, 117], [118, 124], [125, 127], [128, 130], [131, 145], [146, 149], [150, 157], [158, 159], [160, 174], [175, 177], [178, 180], [180, 188], [189, 199], [199, 200], [201, 208], [209, 216], [216, 217], [218, 232], [233, 239], [239, 240], [241, 255], [256, 258], [259, 266], [267, 270], [271, 277], [277, 278], [278, 281], [282, 288], [289, 292], [293, 298], [299, 301], [302, 309], [310, 312], [313, 315], [315, 323], [324, 332], [332, 333]]}
{"doc_key": "ai-train-10", "ner": [[1, 3, "organisation"], [6, 10, "organisation"], [12, 12, "organisation"], [16, 20, "organisation"], [23, 27, "organisation"], [30, 34, "organisation"], [37, 41, "organisation"], [43, 43, "organisation"], [47, 52, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "relations": [[6, 10, 1, 3, "part-of", "", false, false], [12, 12, 6, 10, "named", "", false, false], [16, 20, 1, 3, "part-of", "", false, false], [23, 27, 1, 3, "part-of", "", false, false], [30, 34, 1, 3, "part-of", "", false, false], [37, 41, 1, 3, "part-of", "", false, false], [43, 43, 37, 41, "named", "", false, false], [47, 52, 1, 3, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "sentence": ["La", "National", "Science", "Foundation", "chapeautait", "la", "National", "Aeronautics", "and", "Space", "Administration", "(", "NASA", ")", ",", "le", "minist\u00e8re", "am\u00e9ricain", "de", "l'", "\u00e9nergie", ",", "le", "minist\u00e8re", "am\u00e9ricain", "du", "commerce", "NIST", ",", "le", "minist\u00e8re", "am\u00e9ricain", "de", "la", "d\u00e9fense", ",", "la", "Defense", "Advanced", "Research", "Projects", "Agency", "(", "DARPA", ")", "et", "l'", "Office", "of", "Naval", "Research", ",", "qui", "coordonnaient", "des", "\u00e9tudes", "pour", "informer", "les", "planificateurs", "strat\u00e9giques", "dans", "leurs", "d\u00e9lib\u00e9rations", "."], "sentence-detokenized": "La National Science Foundation chapeautait la National Aeronautics and Space Administration (NASA), le minist\u00e8re am\u00e9ricain de l'\u00e9nergie, le minist\u00e8re am\u00e9ricain du commerce NIST, le minist\u00e8re am\u00e9ricain de la d\u00e9fense, la Defense Advanced Research Projects Agency (DARPA) et l'Office of Naval Research, qui coordonnaient des \u00e9tudes pour informer les planificateurs strat\u00e9giques dans leurs d\u00e9lib\u00e9rations.", "token2charspan": [[0, 2], [3, 11], [12, 19], [20, 30], [31, 42], [43, 45], [46, 54], [55, 66], [67, 70], [71, 76], [77, 91], [92, 93], [93, 97], [97, 98], [98, 99], [100, 102], [103, 112], [113, 122], [123, 125], [126, 128], [128, 135], [135, 136], [137, 139], [140, 149], [150, 159], [160, 162], [163, 171], [172, 176], [176, 177], [178, 180], [181, 190], [191, 200], [201, 203], [204, 206], [207, 214], [214, 215], [216, 218], [219, 226], [227, 235], [236, 244], [245, 253], [254, 260], [261, 262], [262, 267], [267, 268], [269, 271], [272, 274], [274, 280], [281, 283], [284, 289], [290, 298], [298, 299], [300, 303], [304, 317], [318, 321], [322, 328], [329, 333], [334, 342], [343, 346], [347, 361], [362, 374], [375, 379], [380, 385], [386, 399], [399, 400]]}
{"doc_key": "ai-train-11", "ner": [[8, 12, "metrics"], [13, 14, "algorithm"], [19, 20, "researcher"], [26, 26, "researcher"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[8, 12, 13, 14, "part-of", "", false, false], [19, 20, 26, 26, "related-to", "added_appendix_to_work_of", false, false]], "relations_mapping_to_source": [0, 1], "sentence": ["Une", "m\u00e9thode", "rapide", "de", "calcul", "des", "estimations", "du", "maximum", "de", "vraisemblance", "pour", "le", "mod\u00e8le", "probit", "a", "\u00e9t\u00e9", "propos\u00e9e", "par", "Ronald", "Fisher", "en", "annexe", "du", "travail", "de", "Bliss", "en", "1935", "."], "sentence-detokenized": "Une m\u00e9thode rapide de calcul des estimations du maximum de vraisemblance pour le mod\u00e8le probit a \u00e9t\u00e9 propos\u00e9e par Ronald Fisher en annexe du travail de Bliss en 1935.", "token2charspan": [[0, 3], [4, 11], [12, 18], [19, 21], [22, 28], [29, 32], [33, 44], [45, 47], [48, 55], [56, 58], [59, 72], [73, 77], [78, 80], [81, 87], [88, 94], [95, 96], [97, 100], [101, 109], [110, 113], [114, 120], [121, 127], [128, 130], [131, 137], [138, 140], [141, 148], [149, 151], [152, 157], [158, 160], [161, 165], [165, 166]]}
{"doc_key": "ai-train-12", "ner": [[10, 11, "product"], [14, 15, "product"], [20, 20, "organisation"], [18, 18, "product"], [25, 25, "organisation"], [23, 23, "product"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5], "relations": [[18, 18, 14, 15, "usage", "uses_software", false, false], [18, 18, 20, 20, "artifact", "", false, false], [18, 18, 23, 23, "named", "", false, false], [23, 23, 25, 25, "artifact", "", true, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Plusieurs", "de", "ces", "programmes", "sont", "disponibles", "en", "ligne", ",", "comme", "Google", "Translate", "et", "le", "syst\u00e8me", "SYSTRAN", "qui", "alimente", "BabelFish", "d'", "AltaVista", "(", "devenu", "Babelfish", "de", "Yahoo", "depuis", "le", "9", "mai", "2008", ")", "."], "sentence-detokenized": "Plusieurs de ces programmes sont disponibles en ligne, comme Google Translate et le syst\u00e8me SYSTRAN qui alimente BabelFish d'AltaVista (devenu Babelfish de Yahoo depuis le 9 mai 2008).", "token2charspan": [[0, 9], [10, 12], [13, 16], [17, 27], [28, 32], [33, 44], [45, 47], [48, 53], [53, 54], [55, 60], [61, 67], [68, 77], [78, 80], [81, 83], [84, 91], [92, 99], [100, 103], [104, 112], [113, 122], [123, 125], [125, 134], [135, 136], [136, 142], [143, 152], [153, 155], [156, 161], [162, 168], [169, 171], [172, 173], [174, 177], [178, 182], [182, 183], [183, 184]]}
{"doc_key": "ai-train-13", "ner": [[3, 3, "researcher"], [6, 7, "researcher"], [9, 10, "researcher"], [21, 23, "field"], [27, 28, "misc"], [32, 34, "field"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5], "relations": [[3, 3, 21, 23, "related-to", "", true, false], [3, 3, 27, 28, "related-to", "", true, false], [3, 3, 32, 34, "related-to", "", true, false], [6, 7, 21, 23, "related-to", "", true, false], [6, 7, 27, 28, "related-to", "", true, false], [6, 7, 32, 34, "related-to", "", true, false], [9, 10, 21, 23, "related-to", "", true, false], [9, 10, 27, 28, "related-to", "", true, false], [9, 10, 32, 34, "related-to", "", true, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "sentence": ["En", "2002", ",", "Hutter", ",", "avec", "J\u00fcrgen", "Schmidhuber", "et", "Shane", "Legg", ",", "a", "d\u00e9velopp\u00e9", "et", "publi\u00e9", "une", "th\u00e9orie", "math\u00e9matique", "de", "l'", "intelligence", "g\u00e9n\u00e9rale", "artificielle", "bas\u00e9e", "sur", "des", "agents", "intelligents", "id\u00e9alis\u00e9s", "et", "un", "apprentissage", "par", "renforcement", "motiv\u00e9", "par", "la", "r\u00e9compense", "."], "sentence-detokenized": "En 2002, Hutter, avec J\u00fcrgen Schmidhuber et Shane Legg, a d\u00e9velopp\u00e9 et publi\u00e9 une th\u00e9orie math\u00e9matique de l'intelligence g\u00e9n\u00e9rale artificielle bas\u00e9e sur des agents intelligents id\u00e9alis\u00e9s et un apprentissage par renforcement motiv\u00e9 par la r\u00e9compense.", "token2charspan": [[0, 2], [3, 7], [7, 8], [9, 15], [15, 16], [17, 21], [22, 28], [29, 40], [41, 43], [44, 49], [50, 54], [54, 55], [56, 57], [58, 67], [68, 70], [71, 77], [78, 81], [82, 89], [90, 102], [103, 105], [106, 108], [108, 120], [121, 129], [130, 142], [143, 148], [149, 152], [153, 156], [157, 163], [164, 176], [177, 186], [187, 189], [190, 192], [193, 206], [207, 210], [211, 223], [224, 230], [231, 234], [235, 237], [238, 248], [248, 249]]}
{"doc_key": "ai-train-14", "ner": [[11, 11, "metrics"], [13, 17, "metrics"]], "ner_mapping_to_source": [0, 1], "relations": [[11, 11, 13, 17, "named", "", false, false]], "relations_mapping_to_source": [0], "sentence": ["La", "m\u00e9thode", "la", "plus", "courante", "consiste", "\u00e0", "utiliser", "la", "mesure", "dite", "ROUGE", "(", "Recall-Oriented", "Understudy", "for", "Gisting", "Evaluation", ")", "."], "sentence-detokenized": "La m\u00e9thode la plus courante consiste \u00e0 utiliser la mesure dite ROUGE (Recall-Oriented Understudy for Gisting Evaluation).", "token2charspan": [[0, 2], [3, 10], [11, 13], [14, 18], [19, 27], [28, 36], [37, 38], [39, 47], [48, 50], [51, 57], [58, 62], [63, 68], [69, 70], [70, 85], [86, 96], [97, 100], [101, 108], [109, 119], [119, 120], [120, 121]]}
{"doc_key": "ai-train-15", "ner": [[0, 0, "product"], [21, 21, "programlang"], [23, 23, "programlang"], [25, 26, "researcher"], [28, 29, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[0, 0, 21, 21, "related-to", "", false, false], [0, 0, 23, 23, "related-to", "", false, false], [25, 26, 28, 29, "role", "", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["RapidMiner", "fournit", "des", "sch\u00e9mas", "d'", "apprentissage", ",", "des", "mod\u00e8les", "et", "des", "algorithmes", "et", "peut", "\u00eatre", "\u00e9tendu", "\u00e0", "l'", "aide", "de", "scripts", "R", "et", "Python", ".", "David", "Norris", ",", "Bloor", "Research", ",", "13", "novembre", "2013", "."], "sentence-detokenized": "RapidMiner fournit des sch\u00e9mas d'apprentissage, des mod\u00e8les et des algorithmes et peut \u00eatre \u00e9tendu \u00e0 l'aide de scripts R et Python. David Norris, Bloor Research, 13 novembre 2013.", "token2charspan": [[0, 10], [11, 18], [19, 22], [23, 30], [31, 33], [33, 46], [46, 47], [48, 51], [52, 59], [60, 62], [63, 66], [67, 78], [79, 81], [82, 86], [87, 91], [92, 98], [99, 100], [101, 103], [103, 107], [108, 110], [111, 118], [119, 120], [121, 123], [124, 130], [130, 131], [132, 137], [138, 144], [144, 145], [146, 151], [152, 160], [160, 161], [162, 164], [165, 173], [174, 178], [178, 179]]}
{"doc_key": "ai-train-16", "ner": [[1, 3, "product"], [9, 9, "programlang"], [11, 12, "product"]], "ner_mapping_to_source": [0, 4, 5], "relations": [[1, 3, 11, 12, "related-to", "", true, false], [11, 12, 9, 9, "general-affiliation", "", true, false]], "relations_mapping_to_source": [2, 4], "sentence": ["La", "version", "la", "plus", "r\u00e9cente", ",", "enti\u00e8rement", "bas\u00e9e", "sur", "Java", "(", "Weka", "3", ")", ",", "dont", "le", "d\u00e9veloppement", "a", "d\u00e9but\u00e9", "en", "1997", ",", "est", "aujourd'hui", "utilis\u00e9e", "dans", "de", "nombreux", "domaines", "d'", "application", "diff\u00e9rents", ",", "en", "particulier", "\u00e0", "des", "fins", "\u00e9ducatives", "et", "de", "recherche", "."], "sentence-detokenized": "La version la plus r\u00e9cente, enti\u00e8rement bas\u00e9e sur Java (Weka 3), dont le d\u00e9veloppement a d\u00e9but\u00e9 en 1997, est aujourd'hui utilis\u00e9e dans de nombreux domaines d'application diff\u00e9rents, en particulier \u00e0 des fins \u00e9ducatives et de recherche.", "token2charspan": [[0, 2], [3, 10], [11, 13], [14, 18], [19, 26], [26, 27], [28, 39], [40, 45], [46, 49], [50, 54], [55, 56], [56, 60], [61, 62], [62, 63], [63, 64], [65, 69], [70, 72], [73, 86], [87, 88], [89, 95], [96, 98], [99, 103], [103, 104], [105, 108], [109, 120], [121, 129], [130, 134], [135, 137], [138, 146], [147, 155], [156, 158], [158, 169], [170, 180], [180, 181], [182, 184], [185, 196], [197, 198], [199, 202], [203, 207], [208, 218], [219, 221], [222, 224], [225, 234], [234, 235]]}
{"doc_key": "ai-train-17", "ner": [[0, 1, "product"], [17, 24, "misc"], [28, 31, "misc"], [34, 42, "conference"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[17, 24, 0, 1, "topic", "", false, false], [17, 24, 28, 31, "win-defeat", "", false, false], [28, 31, 34, 42, "temporal", "", true, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["Eurisko", "a", "fait", "de", "nombreuses", "d\u00e9couvertes", "int\u00e9ressantes", "et", "a", "b\u00e9n\u00e9fici\u00e9", "d'", "une", "grande", "reconnaissance", ",", "son", "article", "Heuretics", ":", "Theoretical", "and", "Study", "of", "Heuristic", "Rules", "a", "remport\u00e9", "le", "prix", "du", "meilleur", "article", "\u00e0", "l'", "Association", "for", "the", "Advancement", "of", "Artificial", "Intelligence", "de", "1982", "."], "sentence-detokenized": "Eurisko a fait de nombreuses d\u00e9couvertes int\u00e9ressantes et a b\u00e9n\u00e9fici\u00e9 d'une grande reconnaissance, son article Heuretics : Theoretical and Study of Heuristic Rules a remport\u00e9 le prix du meilleur article \u00e0 l'Association for the Advancement of Artificial Intelligence de 1982.", "token2charspan": [[0, 7], [8, 9], [10, 14], [15, 17], [18, 28], [29, 40], [41, 54], [55, 57], [58, 59], [60, 69], [70, 72], [72, 75], [76, 82], [83, 97], [97, 98], [99, 102], [103, 110], [111, 120], [121, 122], [123, 134], [135, 138], [139, 144], [145, 147], [148, 157], [158, 163], [164, 165], [166, 174], [175, 177], [178, 182], [183, 185], [186, 194], [195, 202], [203, 204], [205, 207], [207, 218], [219, 222], [223, 226], [227, 238], [239, 241], [242, 252], [253, 265], [266, 268], [269, 273], [273, 274]]}
{"doc_key": "ai-train-18", "ner": [[8, 10, "metrics"]], "ner_mapping_to_source": [0], "relations": [], "relations_mapping_to_source": [], "sentence": ["Pour", "tenir", "compte", "des", "entit\u00e9s", "multiples", ",", "une", "perte", "de", "charni\u00e8re", "distincte", "est", "calcul\u00e9e", "pour", "chaque", "capsule", "."], "sentence-detokenized": "Pour tenir compte des entit\u00e9s multiples, une perte de charni\u00e8re distincte est calcul\u00e9e pour chaque capsule.", "token2charspan": [[0, 4], [5, 10], [11, 17], [18, 21], [22, 29], [30, 39], [39, 40], [41, 44], [45, 50], [51, 53], [54, 63], [64, 73], [74, 77], [78, 86], [87, 91], [92, 98], [99, 106], [106, 107]]}
{"doc_key": "ai-train-19", "ner": [[8, 10, "product"], [12, 13, "product"], [15, 16, "product"], [18, 19, "product"], [21, 23, "product"], [32, 33, "product"], [40, 45, "product"], [48, 49, "product"], [51, 52, "product"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "relations": [[8, 10, 32, 33, "type-of", "", false, false], [12, 13, 32, 33, "type-of", "", false, false], [15, 16, 32, 33, "type-of", "", false, false], [18, 19, 32, 33, "type-of", "", false, false], [21, 23, 32, 33, "type-of", "", false, false], [48, 49, 40, 45, "type-of", "", false, false], [51, 52, 40, 45, "type-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "sentence": ["Avec", "l'", "\u00e9mergence", "d'", "assistants", "conversationnels", "tels", "que", "Siri", "d'", "Apple", ",", "Amazon", "Alexa", ",", "Google", "Assistant", ",", "Microsoft", "Cortana", "et", "Bixby", "de", "Samsung", ",", "il", "est", "d\u00e9sormais", "possible", "d'", "acc\u00e9der", "aux", "portails", "vocaux", "via", "des", "appareils", "mobiles", "et", "des", "haut-parleurs", "intelligents", "de", "type", "Far", "Field", "tels", "que", "Amazon", "Echo", "et", "Google", "Home", "."], "sentence-detokenized": "Avec l'\u00e9mergence d'assistants conversationnels tels que Siri d'Apple, Amazon Alexa, Google Assistant, Microsoft Cortana et Bixby de Samsung, il est d\u00e9sormais possible d'acc\u00e9der aux portails vocaux via des appareils mobiles et des haut-parleurs intelligents de type Far Field tels que Amazon Echo et Google Home.", "token2charspan": [[0, 4], [5, 7], [7, 16], [17, 19], [19, 29], [30, 46], [47, 51], [52, 55], [56, 60], [61, 63], [63, 68], [68, 69], [70, 76], [77, 82], [82, 83], [84, 90], [91, 100], [100, 101], [102, 111], [112, 119], [120, 122], [123, 128], [129, 131], [132, 139], [139, 140], [141, 143], [144, 147], [148, 157], [158, 166], [167, 169], [169, 176], [177, 180], [181, 189], [190, 196], [197, 200], [201, 204], [205, 214], [215, 222], [223, 225], [226, 229], [230, 243], [244, 256], [257, 259], [260, 264], [265, 268], [269, 274], [275, 279], [280, 283], [284, 290], [291, 295], [296, 298], [299, 305], [306, 310], [310, 311]]}
{"doc_key": "ai-train-20", "ner": [[3, 4, "field"], [7, 9, "algorithm"], [12, 16, "algorithm"], [19, 21, "algorithm"], [24, 24, "algorithm"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[7, 9, 3, 4, "type-of", "", false, false], [12, 16, 3, 4, "type-of", "", false, false], [19, 21, 3, 4, "type-of", "", false, false], [24, 24, 3, 4, "type-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Des", "exemples", "d'", "apprentissage", "supervis\u00e9", "sont", "le", "classificateur", "Naive", "Bayes", ",", "la", "machine", "\u00e0", "vecteurs", "de", "support", ",", "les", "m\u00e9langes", "de", "gaussiens", "et", "le", "r\u00e9seau", "."], "sentence-detokenized": "Des exemples d'apprentissage supervis\u00e9 sont le classificateur Naive Bayes, la machine \u00e0 vecteurs de support, les m\u00e9langes de gaussiens et le r\u00e9seau.", "token2charspan": [[0, 3], [4, 12], [13, 15], [15, 28], [29, 38], [39, 43], [44, 46], [47, 61], [62, 67], [68, 73], [73, 74], [75, 77], [78, 85], [86, 87], [88, 96], [97, 99], [100, 107], [107, 108], [109, 112], [113, 121], [122, 124], [125, 134], [135, 137], [138, 140], [141, 147], [147, 148]]}
{"doc_key": "ai-train-21", "ner": [[4, 5, "algorithm"], [28, 32, "algorithm"], [35, 35, "task"], [40, 41, "metrics"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[4, 5, 28, 32, "part-of", "", true, false], [40, 41, 35, 35, "usage", "", true, false]], "relations_mapping_to_source": [0, 1], "sentence": ["On", "peut", "utiliser", "l'", "algorithme", "OSD", "pour", "d\u00e9river", "des", "limites", "de", "regret", "math\u00e9matique", "O", "(", "\\", "sqrt", "{T}", ")", "/", "math", "pour", "la", "version", "en", "ligne", "de", "la", "machine", "\u00e0", "vecteurs", "de", "support", "pour", "la", "classification", ",", "qui", "utilise", "la", "perte", "charni\u00e8re", "math", "v", "_t", "(", "w", ")", "=", "\\", "max", "\\", "{0", ",", "1", "-", "y", "_t", "(", "w", "\\", "cdot", "x", "_t", ")", "\\}", "/", "math"], "sentence-detokenized": "On peut utiliser l'algorithme OSD pour d\u00e9river des limites de regret math\u00e9matique O (\\ sqrt {T}) / math pour la version en ligne de la machine \u00e0 vecteurs de support pour la classification, qui utilise la perte charni\u00e8re math v _t (w) =\\ max\\ {0, 1 - y _t (w\\ cdot x _t)\\} / math", "token2charspan": [[0, 2], [3, 7], [8, 16], [17, 19], [19, 29], [30, 33], [34, 38], [39, 46], [47, 50], [51, 58], [59, 61], [62, 68], [69, 81], [82, 83], [84, 85], [85, 86], [87, 91], [92, 95], [95, 96], [97, 98], [99, 103], [104, 108], [109, 111], [112, 119], [120, 122], [123, 128], [129, 131], [132, 134], [135, 142], [143, 144], [145, 153], [154, 156], [157, 164], [165, 169], [170, 172], [173, 187], [187, 188], [189, 192], [193, 200], [201, 203], [204, 209], [210, 219], [220, 224], [225, 226], [227, 229], [230, 231], [231, 232], [232, 233], [234, 235], [235, 236], [237, 240], [240, 241], [242, 244], [244, 245], [246, 247], [248, 249], [250, 251], [252, 254], [255, 256], [256, 257], [257, 258], [259, 263], [264, 265], [266, 268], [268, 269], [269, 271], [272, 273], [274, 278]]}
{"doc_key": "ai-train-22", "ner": [[4, 6, "task"], [9, 13, "task"], [12, 12, "task"], [16, 18, "task"], [21, 22, "task"], [25, 27, "task"], [30, 31, "task"], [34, 38, "task"], [41, 43, "task"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "relations": [], "relations_mapping_to_source": [], "sentence": ["Les", "applications", "comprennent", "la", "reconnaissance", "d'", "objets", ",", "la", "cartographie", "et", "la", "navigation", "robotiques", ",", "l'", "assemblage", "d'", "images", ",", "la", "mod\u00e9lisation", "3D", ",", "la", "reconnaissance", "de", "gestes", ",", "le", "suivi", "vid\u00e9o", ",", "l'", "identification", "individuelle", "d'", "animaux", "sauvages", "et", "le", "d\u00e9placement", "d'", "allumettes", "."], "sentence-detokenized": "Les applications comprennent la reconnaissance d'objets, la cartographie et la navigation robotiques, l'assemblage d'images, la mod\u00e9lisation 3D, la reconnaissance de gestes, le suivi vid\u00e9o, l'identification individuelle d'animaux sauvages et le d\u00e9placement d'allumettes.", "token2charspan": [[0, 3], [4, 16], [17, 28], [29, 31], [32, 46], [47, 49], [49, 55], [55, 56], [57, 59], [60, 72], [73, 75], [76, 78], [79, 89], [90, 100], [100, 101], [102, 104], [104, 114], [115, 117], [117, 123], [123, 124], [125, 127], [128, 140], [141, 143], [143, 144], [145, 147], [148, 162], [163, 165], [166, 172], [172, 173], [174, 176], [177, 182], [183, 188], [188, 189], [190, 192], [192, 206], [207, 219], [220, 222], [222, 229], [230, 238], [239, 241], [242, 244], [245, 256], [257, 259], [259, 269], [269, 270]]}
{"doc_key": "ai-train-23", "ner": [[13, 16, "task"], [23, 24, "university"], [28, 30, "university"], [33, 34, "university"], [38, 39, "university"], [43, 48, "university"], [52, 54, "university"], [58, 60, "university"], [64, 65, "university"], [69, 76, "university"], [78, 78, "university"], [83, 87, "university"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11], "relations": [[13, 16, 23, 24, "related-to", "", true, false], [13, 16, 28, 30, "related-to", "", true, false], [13, 16, 33, 34, "related-to", "", true, false], [13, 16, 38, 39, "related-to", "", true, false], [13, 16, 43, 48, "related-to", "", true, false], [13, 16, 52, 54, "related-to", "", true, false], [13, 16, 58, 60, "related-to", "", true, false], [13, 16, 64, 65, "related-to", "", true, false], [13, 16, 69, 76, "related-to", "", true, false], [13, 16, 78, 78, "related-to", "", true, false], [13, 16, 83, 87, "related-to", "", true, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10], "sentence": ["Un", "certain", "nombre", "de", "groupes", "et", "d'", "entreprises", "effectuent", "des", "recherches", "sur", "l'", "estimation", "de", "la", "pose", ",", "notamment", "des", "groupes", "de", "l'", "universit\u00e9", "Brown", ",", "de", "l'", "universit\u00e9", "Carnegie", "Mellon", ",", "du", "MPI", "Saarbruecken", ",", "de", "l'", "universit\u00e9", "Stanford", ",", "de", "l'", "universit\u00e9", "de", "Californie", ",", "San", "Diego", ",", "de", "l'", "universit\u00e9", "de", "Toronto", ",", "de", "l'", "\u00c9cole", "Centrale", "Paris", ",", "de", "l'", "ETH", "Zurich", ",", "de", "l'", "universit\u00e9", "nationale", "des", "sciences", "et", "de", "la", "technologie", "(", "NUST", ")", "et", "de", "l'", "universit\u00e9", "de", "Californie", ",", "Irvine", "."], "sentence-detokenized": "Un certain nombre de groupes et d'entreprises effectuent des recherches sur l'estimation de la pose, notamment des groupes de l'universit\u00e9 Brown, de l'universit\u00e9 Carnegie Mellon, du MPI Saarbruecken, de l'universit\u00e9 Stanford, de l'universit\u00e9 de Californie, San Diego, de l'universit\u00e9 de Toronto, de l'\u00c9cole Centrale Paris, de l'ETH Zurich, de l'universit\u00e9 nationale des sciences et de la technologie (NUST) et de l'universit\u00e9 de Californie, Irvine.", "token2charspan": [[0, 2], [3, 10], [11, 17], [18, 20], [21, 28], [29, 31], [32, 34], [34, 45], [46, 56], [57, 60], [61, 71], [72, 75], [76, 78], [78, 88], [89, 91], [92, 94], [95, 99], [99, 100], [101, 110], [111, 114], [115, 122], [123, 125], [126, 128], [128, 138], [139, 144], [144, 145], [146, 148], [149, 151], [151, 161], [162, 170], [171, 177], [177, 178], [179, 181], [182, 185], [186, 198], [198, 199], [200, 202], [203, 205], [205, 215], [216, 224], [224, 225], [226, 228], [229, 231], [231, 241], [242, 244], [245, 255], [255, 256], [257, 260], [261, 266], [266, 267], [268, 270], [271, 273], [273, 283], [284, 286], [287, 294], [294, 295], [296, 298], [299, 301], [301, 306], [307, 315], [316, 321], [321, 322], [323, 325], [326, 328], [328, 331], [332, 338], [338, 339], [340, 342], [343, 345], [345, 355], [356, 365], [366, 369], [370, 378], [379, 381], [382, 384], [385, 387], [388, 399], [400, 401], [401, 405], [405, 406], [407, 409], [410, 412], [413, 415], [415, 425], [426, 428], [429, 439], [439, 440], [441, 447], [447, 448]]}
{"doc_key": "ai-train-24", "ner": [[0, 6, "metrics"]], "ner_mapping_to_source": [0], "relations": [], "relations_mapping_to_source": [], "sentence": ["La", "fonction", "Sigmo\u00efde", "Perte", "d'", "entropie", "crois\u00e9e", "est", "utilis\u00e9e", "pour", "pr\u00e9dire", "K", "valeurs", "de", "probabilit\u00e9", "ind\u00e9pendantes", "en", "maths", "0,1", "/", "maths", "."], "sentence-detokenized": "La fonction Sigmo\u00efde Perte d'entropie crois\u00e9e est utilis\u00e9e pour pr\u00e9dire K valeurs de probabilit\u00e9 ind\u00e9pendantes en maths 0,1 / maths.", "token2charspan": [[0, 2], [3, 11], [12, 20], [21, 26], [27, 29], [29, 37], [38, 45], [46, 49], [50, 58], [59, 63], [64, 71], [72, 73], [74, 81], [82, 84], [85, 96], [97, 110], [111, 113], [114, 119], [120, 123], [124, 125], [126, 131], [131, 132]]}
{"doc_key": "ai-train-25", "ner": [[6, 8, "misc"], [10, 10, "field"], [13, 13, "field"], [16, 18, "university"], [21, 21, "country"], [26, 28, "misc"], [31, 35, "university"], [38, 38, "country"], [45, 45, "university"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "relations": [[6, 8, 10, 10, "topic", "", false, false], [6, 8, 13, 13, "topic", "", false, false], [6, 8, 16, 18, "physical", "", true, false], [16, 18, 21, 21, "physical", "", false, false], [26, 28, 31, 35, "physical", "", true, false], [31, 35, 38, 38, "physical", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5], "sentence": ["Il", "a", "\u00e9t\u00e9", "titulaire", "de", "la", "chaire", "Johann", "Bernoulli", "de", "math\u00e9matiques", "et", "d'", "informatique", "\u00e0", "l'", "universit\u00e9", "de", "Groningue", ",", "aux", "Pays-Bas", ",", "et", "de", "la", "chaire", "Toshiba", "Endowed", "\u00e0", "l'", "institut", "de", "technologie", "de", "Tokyo", ",", "au", "Japon", ",", "avant", "de", "devenir", "professeur", "\u00e0", "Cambridge", "."], "sentence-detokenized": "Il a \u00e9t\u00e9 titulaire de la chaire Johann Bernoulli de math\u00e9matiques et d'informatique \u00e0 l'universit\u00e9 de Groningue, aux Pays-Bas, et de la chaire Toshiba Endowed \u00e0 l'institut de technologie de Tokyo, au Japon, avant de devenir professeur \u00e0 Cambridge.", "token2charspan": [[0, 2], [3, 4], [5, 8], [9, 18], [19, 21], [22, 24], [25, 31], [32, 38], [39, 48], [49, 51], [52, 65], [66, 68], [69, 71], [71, 83], [84, 85], [86, 88], [88, 98], [99, 101], [102, 111], [111, 112], [113, 116], [117, 125], [125, 126], [127, 129], [130, 132], [133, 135], [136, 142], [143, 150], [151, 158], [159, 160], [161, 163], [163, 171], [172, 174], [175, 186], [187, 189], [190, 195], [195, 196], [197, 199], [200, 205], [205, 206], [207, 212], [213, 215], [216, 223], [224, 234], [235, 236], [237, 246], [246, 247]]}
{"doc_key": "ai-train-26", "ner": [[7, 9, "algorithm"], [14, 18, "algorithm"], [20, 20, "algorithm"], [25, 26, "researcher"], [28, 29, "researcher"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[7, 9, 14, 18, "usage", "", true, false], [14, 18, 25, 26, "origin", "", false, false], [14, 18, 28, 29, "origin", "", false, false], [20, 20, 14, 18, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Une", "autre", "technique", "particuli\u00e8rement", "utilis\u00e9e", "pour", "les", "r\u00e9seaux", "neuronaux", "r\u00e9currents", "est", "le", "r\u00e9seau", "\u00e0", "m\u00e9moire", "\u00e0", "long", "court", "terme", "(", "LSTM", ")", "de", "1997", "par", "Sepp", "Hochreiter", "&", "J\u00fcrgen", "Schmidhuber", "."], "sentence-detokenized": "Une autre technique particuli\u00e8rement utilis\u00e9e pour les r\u00e9seaux neuronaux r\u00e9currents est le r\u00e9seau \u00e0 m\u00e9moire \u00e0 long court terme (LSTM) de 1997 par Sepp Hochreiter & J\u00fcrgen Schmidhuber.", "token2charspan": [[0, 3], [4, 9], [10, 19], [20, 36], [37, 45], [46, 50], [51, 54], [55, 62], [63, 72], [73, 83], [84, 87], [88, 90], [91, 97], [98, 99], [100, 107], [108, 109], [110, 114], [115, 120], [121, 126], [127, 128], [128, 132], [132, 133], [134, 136], [137, 141], [142, 145], [146, 150], [151, 161], [162, 163], [164, 170], [171, 182], [182, 183]]}
{"doc_key": "ai-train-27", "ner": [[5, 7, "programlang"], [9, 9, "product"], [16, 16, "product"], [49, 49, "product"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[9, 9, 5, 7, "general-affiliation", "", false, false], [9, 9, 16, 16, "named", "", false, false]], "relations_mapping_to_source": [0, 1], "sentence": ["L'", "inclusion", "d'", "un", "interpr\u00e9teur", "C", "+", "+", "(", "CINT", "jusqu'", "\u00e0", "la", "version", "5.34", ",", "Cling", "\u00e0", "partir", "de", "la", "version", "6", ")", "rend", "ce", "paquet", "tr\u00e8s", "polyvalent", "car", "il", "peut", "\u00eatre", "utilis\u00e9", "en", "mode", "interactif", ",", "script\u00e9", "et", "compil\u00e9", "d'", "une", "mani\u00e8re", "similaire", "aux", "produits", "commerciaux", "comme", "MATLAB", "."], "sentence-detokenized": "L'inclusion d'un interpr\u00e9teur C + + (CINT jusqu'\u00e0 la version 5.34, Cling \u00e0 partir de la version 6) rend ce paquet tr\u00e8s polyvalent car il peut \u00eatre utilis\u00e9 en mode interactif, script\u00e9 et compil\u00e9 d'une mani\u00e8re similaire aux produits commerciaux comme MATLAB.", "token2charspan": [[0, 2], [2, 11], [12, 14], [14, 16], [17, 29], [30, 31], [32, 33], [34, 35], [36, 37], [37, 41], [42, 48], [48, 49], [50, 52], [53, 60], [61, 65], [65, 66], [67, 72], [73, 74], [75, 81], [82, 84], [85, 87], [88, 95], [96, 97], [97, 98], [99, 103], [104, 106], [107, 113], [114, 118], [119, 129], [130, 133], [134, 136], [137, 141], [142, 146], [147, 154], [155, 157], [158, 162], [163, 173], [173, 174], [175, 182], [183, 185], [186, 193], [194, 196], [196, 199], [200, 207], [208, 217], [218, 221], [222, 230], [231, 242], [243, 248], [249, 255], [255, 256]]}
{"doc_key": "ai-train-28", "ner": [[0, 2, "product"], [29, 32, "field"], [36, 38, "task"], [41, 44, "task"], [47, 49, "task"], [52, 54, "task"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5], "relations": [[0, 2, 29, 32, "related-to", "", false, false], [36, 38, 29, 32, "part-of", "", false, false], [41, 44, 29, 32, "part-of", "", false, false], [47, 49, 29, 32, "part-of", "", false, false], [52, 54, 29, 32, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4], "sentence": ["Les", "interfaces", "vocales", "qui", "interpr\u00e8tent", "et", "g\u00e8rent", "l'", "\u00e9tat", "de", "la", "conversation", "sont", "difficiles", "\u00e0", "concevoir", "en", "raison", "de", "la", "difficult\u00e9", "inh\u00e9rente", "\u00e0", "l'", "int\u00e9gration", "de", "t\u00e2ches", "complexes", "de", "traitement", "du", "langage", "naturel", "telles", "que", "la", "r\u00e9solution", "de", "cor\u00e9f\u00e9rences", ",", "la", "reconnaissance", "d'", "entit\u00e9s", "nomm\u00e9es", ",", "la", "recherche", "d'", "informations", "et", "la", "gestion", "du", "dialogue", "."], "sentence-detokenized": "Les interfaces vocales qui interpr\u00e8tent et g\u00e8rent l'\u00e9tat de la conversation sont difficiles \u00e0 concevoir en raison de la difficult\u00e9 inh\u00e9rente \u00e0 l'int\u00e9gration de t\u00e2ches complexes de traitement du langage naturel telles que la r\u00e9solution de cor\u00e9f\u00e9rences, la reconnaissance d'entit\u00e9s nomm\u00e9es, la recherche d'informations et la gestion du dialogue.", "token2charspan": [[0, 3], [4, 14], [15, 22], [23, 26], [27, 39], [40, 42], [43, 49], [50, 52], [52, 56], [57, 59], [60, 62], [63, 75], [76, 80], [81, 91], [92, 93], [94, 103], [104, 106], [107, 113], [114, 116], [117, 119], [120, 130], [131, 140], [141, 142], [143, 145], [145, 156], [157, 159], [160, 166], [167, 176], [177, 179], [180, 190], [191, 193], [194, 201], [202, 209], [210, 216], [217, 220], [221, 223], [224, 234], [235, 237], [238, 250], [250, 251], [252, 254], [255, 269], [270, 272], [272, 279], [280, 287], [287, 288], [289, 291], [292, 301], [302, 304], [304, 316], [317, 319], [320, 322], [323, 330], [331, 333], [334, 342], [342, 343]]}
{"doc_key": "ai-train-29", "ner": [[6, 8, "algorithm"], [11, 14, "algorithm"], [22, 23, "researcher"], [25, 28, "organisation"], [35, 37, "field"], [40, 41, "field"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5], "relations": [[6, 8, 22, 23, "origin", "", false, false], [6, 8, 35, 37, "part-of", "", false, false], [6, 8, 40, 41, "part-of", "", false, false], [11, 14, 22, 23, "origin", "", false, false], [11, 14, 35, 37, "part-of", "", false, false], [11, 14, 40, 41, "part-of", "", false, false], [22, 23, 25, 28, "physical", "", false, false], [22, 23, 25, 28, "role", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "sentence": ["Entre", "2009", "et", "2012", ",", "les", "r\u00e9seaux", "neuronaux", "r\u00e9currents", "et", "les", "r\u00e9seaux", "neuronaux", "feedforward", "profonds", "d\u00e9velopp\u00e9s", "par", "le", "groupe", "de", "recherche", "de", "J\u00fcrgen", "Schmidhuber", "au", "Swiss", "AI", "Lab", "IDSIA", "ont", "remport\u00e9", "huit", "concours", "internationaux", "de", "reconnaissance", "de", "formes", "et", "d'", "apprentissage", "automatique", "."], "sentence-detokenized": "Entre 2009 et 2012, les r\u00e9seaux neuronaux r\u00e9currents et les r\u00e9seaux neuronaux feedforward profonds d\u00e9velopp\u00e9s par le groupe de recherche de J\u00fcrgen Schmidhuber au Swiss AI Lab IDSIA ont remport\u00e9 huit concours internationaux de reconnaissance de formes et d'apprentissage automatique.", "token2charspan": [[0, 5], [6, 10], [11, 13], [14, 18], [18, 19], [20, 23], [24, 31], [32, 41], [42, 52], [53, 55], [56, 59], [60, 67], [68, 77], [78, 89], [90, 98], [99, 109], [110, 113], [114, 116], [117, 123], [124, 126], [127, 136], [137, 139], [140, 146], [147, 158], [159, 161], [162, 167], [168, 170], [171, 174], [175, 180], [181, 184], [185, 193], [194, 198], [199, 207], [208, 222], [223, 225], [226, 240], [241, 243], [244, 250], [251, 253], [254, 256], [256, 269], [270, 281], [281, 282]]}
{"doc_key": "ai-train-30", "ner": [[0, 4, "product"], [10, 11, "product"], [13, 14, "product"], [20, 21, "task"], [24, 24, "task"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[0, 4, 10, 11, "usage", "", false, false], [0, 4, 13, 14, "usage", "", false, false], [0, 4, 20, 21, "usage", "", true, false], [0, 4, 24, 24, "usage", "", true, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Les", "syst\u00e8mes", "de", "bureau", "Windows", "modernes", "peuvent", "utiliser", "les", "composants", "SAPI", "4", "et", "SAPI", "5", "pour", "prendre", "en", "charge", "la", "synth\u00e8se", "vocale", "et", "la", "parole", "."], "sentence-detokenized": "Les syst\u00e8mes de bureau Windows modernes peuvent utiliser les composants SAPI 4 et SAPI 5 pour prendre en charge la synth\u00e8se vocale et la parole.", "token2charspan": [[0, 3], [4, 12], [13, 15], [16, 22], [23, 30], [31, 39], [40, 47], [48, 56], [57, 60], [61, 71], [72, 76], [77, 78], [79, 81], [82, 86], [87, 88], [89, 93], [94, 101], [102, 104], [105, 111], [112, 114], [115, 123], [124, 130], [131, 133], [134, 136], [137, 143], [143, 144]]}
{"doc_key": "ai-train-31", "ner": [[8, 13, "misc"], [15, 15, "field"], [18, 20, "university"], [27, 30, "field"], [33, 37, "university"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[8, 13, 15, 15, "topic", "topic_of_award", false, false], [8, 13, 18, 20, "origin", "", true, false], [27, 30, 33, 37, "related-to", "", true, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["Il", "a", "re\u00e7u", "deux", "dipl\u00f4mes", "honorifiques", ",", "un", "S.", "V.", "della", "laurea", "ad", "honorem", "en", "psychologie", "de", "l'", "universit\u00e9", "de", "Padoue", "en", "1995", "et", "un", "doctorat", "en", "conception", "et", "ing\u00e9nierie", "industrielles", "de", "l'", "universit\u00e9", "de", "technologie", "de", "Delft", "."], "sentence-detokenized": "Il a re\u00e7u deux dipl\u00f4mes honorifiques, un S. V. della laurea ad honorem en psychologie de l'universit\u00e9 de Padoue en 1995 et un doctorat en conception et ing\u00e9nierie industrielles de l'universit\u00e9 de technologie de Delft.", "token2charspan": [[0, 2], [3, 4], [5, 9], [10, 14], [15, 23], [24, 36], [36, 37], [38, 40], [41, 43], [44, 46], [47, 52], [53, 59], [60, 62], [63, 70], [71, 73], [74, 85], [86, 88], [89, 91], [91, 101], [102, 104], [105, 111], [112, 114], [115, 119], [120, 122], [123, 125], [126, 134], [135, 137], [138, 148], [149, 151], [152, 162], [163, 176], [177, 179], [180, 182], [182, 192], [193, 195], [196, 207], [208, 210], [211, 216], [216, 217]]}
{"doc_key": "ai-train-32", "ner": [[7, 8, "researcher"], [13, 16, "organisation"], [18, 18, "location"], [20, 21, "researcher"], [33, 34, "misc"], [50, 52, "misc"], [72, 73, "misc"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "relations": [[7, 8, 13, 16, "physical", "", false, false], [7, 8, 13, 16, "role", "", false, false], [13, 16, 18, 18, "physical", "", false, false], [20, 21, 33, 34, "related-to", "works_with", true, false], [20, 21, 50, 52, "related-to", "works_with", true, false], [20, 21, 72, 73, "related-to", "works_with", true, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5], "sentence": ["Avec", "son", "collaborateur", "de", "longue", "date", ",", "Laurent", "Cohen", ",", "neurologue", "\u00e0", "l'", "h\u00f4pital", "de", "la", "Piti\u00e9-Salp\u00eatri\u00e8re", "\u00e0", "Paris", ",", "Dehaene", "a", "\u00e9galement", "identifi\u00e9", "des", "patients", "pr\u00e9sentant", "des", "l\u00e9sions", "dans", "diff\u00e9rentes", "r\u00e9gions", "du", "lobe", "pari\u00e9tal", "avec", "une", "multiplication", "alt\u00e9r\u00e9e", ",", "mais", "une", "soustraction", "pr\u00e9serv\u00e9e", "(", "associ\u00e9e", "\u00e0", "des", "l\u00e9sions", "du", "lobule", "pari\u00e9tal", "inf\u00e9rieur", ")", "et", "d'", "autres", "avec", "une", "soustraction", "alt\u00e9r\u00e9e", ",", "mais", "une", "multiplication", "pr\u00e9serv\u00e9e", "(", "associ\u00e9e", "\u00e0", "des", "l\u00e9sions", "du", "sillon", "intrapari\u00e9tal", ")", "."], "sentence-detokenized": "Avec son collaborateur de longue date, Laurent Cohen, neurologue \u00e0 l'h\u00f4pital de la Piti\u00e9-Salp\u00eatri\u00e8re \u00e0 Paris, Dehaene a \u00e9galement identifi\u00e9 des patients pr\u00e9sentant des l\u00e9sions dans diff\u00e9rentes r\u00e9gions du lobe pari\u00e9tal avec une multiplication alt\u00e9r\u00e9e, mais une soustraction pr\u00e9serv\u00e9e (associ\u00e9e \u00e0 des l\u00e9sions du lobule pari\u00e9tal inf\u00e9rieur) et d'autres avec une soustraction alt\u00e9r\u00e9e, mais une multiplication pr\u00e9serv\u00e9e (associ\u00e9e \u00e0 des l\u00e9sions du sillon intrapari\u00e9tal).", "token2charspan": [[0, 4], [5, 8], [9, 22], [23, 25], [26, 32], [33, 37], [37, 38], [39, 46], [47, 52], [52, 53], [54, 64], [65, 66], [67, 69], [69, 76], [77, 79], [80, 82], [83, 100], [101, 102], [103, 108], [108, 109], [110, 117], [118, 119], [120, 129], [130, 139], [140, 143], [144, 152], [153, 163], [164, 167], [168, 175], [176, 180], [181, 192], [193, 200], [201, 203], [204, 208], [209, 217], [218, 222], [223, 226], [227, 241], [242, 249], [249, 250], [251, 255], [256, 259], [260, 272], [273, 282], [283, 284], [284, 292], [293, 294], [295, 298], [299, 306], [307, 309], [310, 316], [317, 325], [326, 335], [335, 336], [337, 339], [340, 342], [342, 348], [349, 353], [354, 357], [358, 370], [371, 378], [378, 379], [380, 384], [385, 388], [389, 403], [404, 413], [414, 415], [415, 423], [424, 425], [426, 429], [430, 437], [438, 440], [441, 447], [448, 461], [461, 462], [462, 463]]}
{"doc_key": "ai-train-33", "ner": [[7, 9, "product"], [15, 17, "misc"], [19, 20, "misc"], [28, 28, "misc"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[15, 17, 7, 9, "topic", "", false, false], [19, 20, 7, 9, "topic", "", false, false], [28, 28, 7, 9, "topic", "", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["Plus", "r\u00e9cemment", ",", "les", "repr\u00e9sentations", "fictionnelles", "de", "robots", "artificiellement", "intelligents", "dans", "des", "films", "tels", "que", "A.I.", "Artificial", "Intelligence", "et", "Ex", "Machina", "et", "l'", "adaptation", "t\u00e9l\u00e9vis\u00e9e", "de", "2016", "de", "Westworld", "ont", "suscit\u00e9", "la", "sympathie", "du", "public", "pour", "les", "robots", "eux-m\u00eames", "."], "sentence-detokenized": "Plus r\u00e9cemment, les repr\u00e9sentations fictionnelles de robots artificiellement intelligents dans des films tels que A.I. Artificial Intelligence et Ex Machina et l'adaptation t\u00e9l\u00e9vis\u00e9e de 2016 de Westworld ont suscit\u00e9 la sympathie du public pour les robots eux-m\u00eames.", "token2charspan": [[0, 4], [5, 14], [14, 15], [16, 19], [20, 35], [36, 49], [50, 52], [53, 59], [60, 76], [77, 89], [90, 94], [95, 98], [99, 104], [105, 109], [110, 113], [114, 118], [119, 129], [130, 142], [143, 145], [146, 148], [149, 156], [157, 159], [160, 162], [162, 172], [173, 182], [183, 185], [186, 190], [191, 193], [194, 203], [204, 207], [208, 215], [216, 218], [219, 228], [229, 231], [232, 238], [239, 243], [244, 247], [248, 254], [255, 264], [264, 265]]}
{"doc_key": "ai-train-34", "ner": [[7, 9, "field"], [12, 15, "algorithm"], [18, 20, "task"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[12, 15, 7, 9, "part-of", "", false, false], [18, 20, 7, 9, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1], "sentence": ["Deux", "des", "principales", "m\u00e9thodes", "utilis\u00e9es", "dans", "l'", "apprentissage", "non", "supervis\u00e9", "sont", "l'", "analyse", "en", "composantes", "principales", "et", "l'", "analyse", "en", "grappes", "."], "sentence-detokenized": "Deux des principales m\u00e9thodes utilis\u00e9es dans l'apprentissage non supervis\u00e9 sont l'analyse en composantes principales et l'analyse en grappes.", "token2charspan": [[0, 4], [5, 8], [9, 20], [21, 29], [30, 39], [40, 44], [45, 47], [47, 60], [61, 64], [65, 74], [75, 79], [80, 82], [82, 89], [90, 92], [93, 104], [105, 116], [117, 119], [120, 122], [122, 129], [130, 132], [133, 140], [140, 141]]}
{"doc_key": "ai-train-35", "ner": [[0, 3, "organisation"], [27, 28, "misc"], [33, 34, "misc"], [36, 38, "person"], [43, 44, "person"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[27, 28, 0, 3, "artifact", "", false, false], [33, 34, 0, 3, "artifact", "", false, false], [33, 34, 36, 38, "role", "director_of", false, false], [33, 34, 43, 44, "role", "actor_in", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["La", "Walt", "Disney", "Company", "a", "\u00e9galement", "commenc\u00e9", "\u00e0", "utiliser", "de", "mani\u00e8re", "plus", "importante", "les", "films", "en", "3D", "dans", "des", "lieux", "sp\u00e9ciaux", "pour", "impressionner", "le", "public", ".", "Les", "Voyages", "magiques", "(", "1982", ")", "et", "Captain", "EO", "(", "Francis", "Ford", "Coppola", ",", "1986", ",", "avec", "Michael", "Jackson", ")", "en", "sont", "des", "exemples", "notables", "."], "sentence-detokenized": "La Walt Disney Company a \u00e9galement commenc\u00e9 \u00e0 utiliser de mani\u00e8re plus importante les films en 3D dans des lieux sp\u00e9ciaux pour impressionner le public. Les Voyages magiques (1982) et Captain EO (Francis Ford Coppola, 1986, avec Michael Jackson) en sont des exemples notables.", "token2charspan": [[0, 2], [3, 7], [8, 14], [15, 22], [23, 24], [25, 34], [35, 43], [44, 45], [46, 54], [55, 57], [58, 65], [66, 70], [71, 81], [82, 85], [86, 91], [92, 94], [95, 97], [98, 102], [103, 106], [107, 112], [113, 121], [122, 126], [127, 140], [141, 143], [144, 150], [150, 151], [152, 155], [156, 163], [164, 172], [173, 174], [174, 178], [178, 179], [180, 182], [183, 190], [191, 193], [194, 195], [195, 202], [203, 207], [208, 215], [215, 216], [217, 221], [221, 222], [223, 227], [228, 235], [236, 243], [243, 244], [245, 247], [248, 252], [253, 256], [257, 265], [266, 274], [274, 275]]}
{"doc_key": "ai-train-36", "ner": [[14, 17, "field"], [24, 28, "task"], [31, 32, "task"], [34, 34, "researcher"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[24, 28, 14, 17, "part-of", "", false, false], [31, 32, 14, 17, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1], "sentence": ["Depuis", "2002", ",", "l'", "entra\u00eenement", "par", "perceptron", "est", "devenu", "populaire", "dans", "le", "domaine", "du", "traitement", "du", "langage", "naturel", "pour", "des", "t\u00e2ches", "telles", "que", "l'", "\u00e9tiquetage", "des", "parties", "du", "discours", "et", "l'", "analyse", "syntaxique", "(", "Collins", ",", "2002", ")", "."], "sentence-detokenized": "Depuis 2002, l'entra\u00eenement par perceptron est devenu populaire dans le domaine du traitement du langage naturel pour des t\u00e2ches telles que l'\u00e9tiquetage des parties du discours et l'analyse syntaxique (Collins, 2002).", "token2charspan": [[0, 6], [7, 11], [11, 12], [13, 15], [15, 27], [28, 31], [32, 42], [43, 46], [47, 53], [54, 63], [64, 68], [69, 71], [72, 79], [80, 82], [83, 93], [94, 96], [97, 104], [105, 112], [113, 117], [118, 121], [122, 128], [129, 135], [136, 139], [140, 142], [142, 152], [153, 156], [157, 164], [165, 167], [168, 176], [177, 179], [180, 182], [182, 189], [190, 200], [201, 202], [202, 209], [209, 210], [211, 215], [215, 216], [216, 217]]}
{"doc_key": "ai-train-37", "ner": [[2, 4, "product"], [12, 16, "organisation"], [18, 19, "organisation"], [21, 21, "country"], [25, 30, "product"], [35, 36, "researcher"], [47, 47, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "relations": [[12, 16, 2, 4, "role", "introduces_to_market", true, false], [18, 19, 2, 4, "role", "introduces_to_market", true, false], [18, 19, 21, 21, "physical", "", false, false], [25, 30, 47, 47, "related-to", "sold_to", true, false], [35, 36, 25, 30, "origin", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4], "sentence": ["Le", "premier", "robot", "de", "palettisation", "a", "\u00e9t\u00e9", "introduit", "en", "1963", "par", "la", "soci\u00e9t\u00e9", "Fuji", "Yusoki", "Kogyo", ".", "par", "KUKA", "robotics", "en", "Allemagne", ",", "et", "la", "machine", "universelle", "programmable", "pour", "l'", "assemblage", "a", "\u00e9t\u00e9", "invent\u00e9e", "par", "Victor", "Scheinman", "en", "1976", ",", "et", "la", "conception", "a", "\u00e9t\u00e9", "vendue", "\u00e0", "Unimation", "."], "sentence-detokenized": "Le premier robot de palettisation a \u00e9t\u00e9 introduit en 1963 par la soci\u00e9t\u00e9 Fuji Yusoki Kogyo. par KUKA robotics en Allemagne, et la machine universelle programmable pour l'assemblage a \u00e9t\u00e9 invent\u00e9e par Victor Scheinman en 1976, et la conception a \u00e9t\u00e9 vendue \u00e0 Unimation.", "token2charspan": [[0, 2], [3, 10], [11, 16], [17, 19], [20, 33], [34, 35], [36, 39], [40, 49], [50, 52], [53, 57], [58, 61], [62, 64], [65, 72], [73, 77], [78, 84], [85, 90], [90, 91], [92, 95], [96, 100], [101, 109], [110, 112], [113, 122], [122, 123], [124, 126], [127, 129], [130, 137], [138, 149], [150, 162], [163, 167], [168, 170], [170, 180], [181, 182], [183, 186], [187, 195], [196, 199], [200, 206], [207, 216], [217, 219], [220, 224], [224, 225], [226, 228], [229, 231], [232, 242], [243, 244], [245, 248], [249, 255], [256, 257], [258, 267], [267, 268]]}
{"doc_key": "ai-train-38", "ner": [[13, 13, "conference"], [15, 16, "researcher"], [27, 27, "field"], [44, 45, "researcher"], [53, 54, "researcher"], [67, 67, "field"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5], "relations": [[15, 16, 13, 13, "role", "president_of", false, false], [15, 16, 44, 45, "role", "colleagues", false, false], [27, 27, 67, 67, "named", "same", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["Au", "milieu", "des", "ann\u00e9es", "1990", ",", "alors", "qu'", "il", "\u00e9tait", "pr\u00e9sident", "de", "l'", "AAAI", ",", "Hayes", "a", "lanc\u00e9", "une", "s\u00e9rie", "d'", "attaques", "contre", "les", "critiques", "de", "l'", "IA", ",", "le", "plus", "souvent", "formul\u00e9es", "de", "mani\u00e8re", "ironique", ",", "et", "a", "invent\u00e9", "(", "avec", "son", "coll\u00e8gue", "Kenneth", "Ford", ")", "un", "prix", "portant", "le", "nom", "de", "Simon", "Newcomb", "et", "r\u00e9compensant", "l'", "argument", "le", "plus", "ridicule", "r\u00e9futant", "la", "possibilit\u00e9", "de", "l'", "IA", "."], "sentence-detokenized": "Au milieu des ann\u00e9es 1990, alors qu'il \u00e9tait pr\u00e9sident de l'AAAI, Hayes a lanc\u00e9 une s\u00e9rie d'attaques contre les critiques de l'IA, le plus souvent formul\u00e9es de mani\u00e8re ironique, et a invent\u00e9 (avec son coll\u00e8gue Kenneth Ford) un prix portant le nom de Simon Newcomb et r\u00e9compensant l'argument le plus ridicule r\u00e9futant la possibilit\u00e9 de l'IA.", "token2charspan": [[0, 2], [3, 9], [10, 13], [14, 20], [21, 25], [25, 26], [27, 32], [33, 36], [36, 38], [39, 44], [45, 54], [55, 57], [58, 60], [60, 64], [64, 65], [66, 71], [72, 73], [74, 79], [80, 83], [84, 89], [90, 92], [92, 100], [101, 107], [108, 111], [112, 121], [122, 124], [125, 127], [127, 129], [129, 130], [131, 133], [134, 138], [139, 146], [147, 156], [157, 159], [160, 167], [168, 176], [176, 177], [178, 180], [181, 182], [183, 190], [191, 192], [192, 196], [197, 200], [201, 209], [210, 217], [218, 222], [222, 223], [224, 226], [227, 231], [232, 239], [240, 242], [243, 246], [247, 249], [250, 255], [256, 263], [264, 266], [267, 279], [280, 282], [282, 290], [291, 293], [294, 298], [299, 307], [308, 316], [317, 319], [320, 331], [332, 334], [335, 337], [337, 339], [339, 340]]}
{"doc_key": "ai-train-39", "ner": [[16, 19, "algorithm"], [46, 47, "algorithm"], [60, 65, "algorithm"], [69, 71, "algorithm"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[16, 19, 46, 47, "named", "same", false, false], [60, 65, 16, 19, "type-of", "", false, false], [69, 71, 16, 19, "type-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["Une", "valeur", "optimale", "de", "math\u00e9matique", "alpha", "/", "math", "peut", "\u00eatre", "trouv\u00e9e", "\u00e0", "l'", "aide", "d'", "un", "algorithme", "de", "recherche", "lin\u00e9aire", ",", "c'est-\u00e0-dire", "que", "la", "valeur", "de", "math\u00e9matique", "alpha", "/", "math", "est", "d\u00e9termin\u00e9e", "en", "trouvant", "la", "valeur", "qui", "minimise", "S", ",", "g\u00e9n\u00e9ralement", "\u00e0", "l'", "aide", "d'", "une", "recherche", "lin\u00e9aire", "dans", "l'", "intervalle", "math0", "\\", "alpha", "1", "/", "math", "ou", "d'", "une", "recherche", "lin\u00e9aire", "par", "retour", "en", "arri\u00e8re", "telle", "que", "la", "recherche", "lin\u00e9aire", "Armijo", "."], "sentence-detokenized": "Une valeur optimale de math\u00e9matique alpha / math peut \u00eatre trouv\u00e9e \u00e0 l'aide d'un algorithme de recherche lin\u00e9aire, c'est-\u00e0-dire que la valeur de math\u00e9matique alpha / math est d\u00e9termin\u00e9e en trouvant la valeur qui minimise S, g\u00e9n\u00e9ralement \u00e0 l'aide d'une recherche lin\u00e9aire dans l'intervalle math0\\ alpha 1 / math ou d'une recherche lin\u00e9aire par retour en arri\u00e8re telle que la recherche lin\u00e9aire Armijo.", "token2charspan": [[0, 3], [4, 10], [11, 19], [20, 22], [23, 35], [36, 41], [42, 43], [44, 48], [49, 53], [54, 58], [59, 66], [67, 68], [69, 71], [71, 75], [76, 78], [78, 80], [81, 91], [92, 94], [95, 104], [105, 113], [113, 114], [115, 127], [128, 131], [132, 134], [135, 141], [142, 144], [145, 157], [158, 163], [164, 165], [166, 170], [171, 174], [175, 185], [186, 188], [189, 197], [198, 200], [201, 207], [208, 211], [212, 220], [221, 222], [222, 223], [224, 236], [237, 238], [239, 241], [241, 245], [246, 248], [248, 251], [252, 261], [262, 270], [271, 275], [276, 278], [278, 288], [289, 294], [294, 295], [296, 301], [302, 303], [304, 305], [306, 310], [311, 313], [314, 316], [316, 319], [320, 329], [330, 338], [339, 342], [343, 349], [350, 352], [353, 360], [361, 366], [367, 370], [371, 373], [374, 383], [384, 392], [393, 399], [399, 400]]}
{"doc_key": "ai-train-40", "ner": [[5, 7, "algorithm"], [9, 10, "algorithm"], [20, 21, "product"]], "ner_mapping_to_source": [0, 1, 2], "relations": [], "relations_mapping_to_source": [], "sentence": ["Il", "aborde", "les", "techniques", "de", "recherche", "en", "largeur", "et", "en", "profondeur", ",", "mais", "conclut", "finalement", "que", "les", "r\u00e9sultats", "repr\u00e9sentent", "des", "syst\u00e8mes", "experts", "qui", "incarnent", "beaucoup", "de", "connaissances", "techniques", ",", "mais", "ne", "font", "pas", "beaucoup", "de", "lumi\u00e8re", "sur", "les", "processus", "mentaux", "que", "les", "humains", "utilisent", "pour", "r\u00e9soudre", "de", "telles", "\u00e9nigmes", "."], "sentence-detokenized": "Il aborde les techniques de recherche en largeur et en profondeur, mais conclut finalement que les r\u00e9sultats repr\u00e9sentent des syst\u00e8mes experts qui incarnent beaucoup de connaissances techniques, mais ne font pas beaucoup de lumi\u00e8re sur les processus mentaux que les humains utilisent pour r\u00e9soudre de telles \u00e9nigmes.", "token2charspan": [[0, 2], [3, 9], [10, 13], [14, 24], [25, 27], [28, 37], [38, 40], [41, 48], [49, 51], [52, 54], [55, 65], [65, 66], [67, 71], [72, 79], [80, 90], [91, 94], [95, 98], [99, 108], [109, 121], [122, 125], [126, 134], [135, 142], [143, 146], [147, 156], [157, 165], [166, 168], [169, 182], [183, 193], [193, 194], [195, 199], [200, 202], [203, 207], [208, 211], [212, 220], [221, 223], [224, 231], [232, 235], [236, 239], [240, 249], [250, 257], [258, 261], [262, 265], [266, 273], [274, 283], [284, 288], [289, 297], [298, 300], [301, 307], [308, 315], [315, 316]]}
{"doc_key": "ai-train-41", "ner": [[0, 5, "task"], [4, 4, "task"]], "ner_mapping_to_source": [0, 1], "relations": [], "relations_mapping_to_source": [], "sentence": ["La", "reconnaissance", "et", "la", "synth\u00e8se", "vocales", "traitent", "de", "la", "mani\u00e8re", "dont", "le", "langage", "parl\u00e9", "peut", "\u00eatre", "compris", "ou", "cr\u00e9\u00e9", "\u00e0", "l'", "aide", "d'", "ordinateurs", "."], "sentence-detokenized": "La reconnaissance et la synth\u00e8se vocales traitent de la mani\u00e8re dont le langage parl\u00e9 peut \u00eatre compris ou cr\u00e9\u00e9 \u00e0 l'aide d'ordinateurs.", "token2charspan": [[0, 2], [3, 17], [18, 20], [21, 23], [24, 32], [33, 40], [41, 49], [50, 52], [53, 55], [56, 63], [64, 68], [69, 71], [72, 79], [80, 85], [86, 90], [91, 95], [96, 103], [104, 106], [107, 111], [112, 113], [114, 116], [116, 120], [121, 123], [123, 134], [134, 135]]}
{"doc_key": "ai-train-42", "ner": [[18, 20, "algorithm"], [35, 37, "algorithm"]], "ner_mapping_to_source": [0, 1], "relations": [], "relations_mapping_to_source": [], "sentence": ["Cette", "math\u00e9matique", "du", "th\u00eata", "^", "{*}", "/", "math", "est", "normalement", "estim\u00e9e", "\u00e0", "l'", "aide", "d'", "une", "proc\u00e9dure", "de", "Maximum", "de", "Vraisemblance", "(", "math\\\\_theta", "^", "{*}", "=", "\\", "theta", "^", "{ML}", "/", "math", ")", "ou", "de", "Maximum", "A", "Posteriori", "(", "math\\_theta", "^", "{*}", "=", "\\theta", "^", "{", "MAP}", "/", "math", ")", "."], "sentence-detokenized": "Cette math\u00e9matique du th\u00eata ^ {*} / math est normalement estim\u00e9e \u00e0 l'aide d'une proc\u00e9dure de Maximum de Vraisemblance (math\\\\_theta ^ {*} =\\theta ^ {ML} / math) ou de Maximum A Posteriori (math\\_theta ^ {*} =\\theta ^ {MAP} / math).", "token2charspan": [[0, 5], [6, 18], [19, 21], [22, 27], [28, 29], [30, 33], [34, 35], [36, 40], [41, 44], [45, 56], [57, 64], [65, 66], [67, 69], [69, 73], [74, 76], [76, 79], [80, 89], [90, 92], [93, 100], [101, 103], [104, 117], [118, 119], [119, 131], [132, 133], [134, 137], [138, 139], [139, 140], [140, 145], [146, 147], [148, 152], [153, 154], [155, 159], [159, 160], [161, 163], [164, 166], [167, 174], [175, 176], [177, 187], [188, 189], [189, 200], [201, 202], [203, 206], [207, 208], [208, 214], [215, 216], [217, 218], [218, 222], [223, 224], [225, 229], [229, 230], [230, 231]]}
{"doc_key": "ai-train-43", "ner": [[6, 8, "product"]], "ner_mapping_to_source": [0], "relations": [], "relations_mapping_to_source": [], "sentence": ["Certaines", "langues", "moins", "r\u00e9pandues", "utilisent", "le", "synth\u00e9tiseur", "open-source", "eSpeak", "pour", "leur", "discours", ",", "produisant", "une", "voix", "robotique", "et", "maladroite", "qui", "peut", "\u00eatre", "difficile", "\u00e0", "comprendre", "."], "sentence-detokenized": "Certaines langues moins r\u00e9pandues utilisent le synth\u00e9tiseur open-source eSpeak pour leur discours, produisant une voix robotique et maladroite qui peut \u00eatre difficile \u00e0 comprendre.", "token2charspan": [[0, 9], [10, 17], [18, 23], [24, 33], [34, 43], [44, 46], [47, 59], [60, 71], [72, 78], [79, 83], [84, 88], [89, 97], [97, 98], [99, 109], [110, 113], [114, 118], [119, 128], [129, 131], [132, 142], [143, 146], [147, 151], [152, 156], [157, 166], [167, 168], [169, 179], [179, 180]]}
{"doc_key": "ai-train-44", "ner": [[25, 25, "programlang"], [47, 48, "programlang"], [50, 50, "product"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[25, 25, 47, 48, "compare", "", false, false], [25, 25, 50, 50, "compare", "", false, false]], "relations_mapping_to_source": [0, 1], "sentence": ["Bien", "qu'", "il", "soit", "principalement", "utilis\u00e9", "par", "les", "statisticiens", "et", "autres", "praticiens", "n\u00e9cessitant", "un", "environnement", "pour", "le", "calcul", "statistique", "et", "le", "d\u00e9veloppement", "de", "logiciels", ",", "R", "peut", "\u00e9galement", "fonctionner", "comme", "une", "bo\u00eete", "\u00e0", "outils", "g\u00e9n\u00e9rale", "pour", "le", "calcul", "matriciel", "-", "avec", "des", "performances", "comparables", "\u00e0", "celles", "de", "GNU", "Octave", "ou", "MATLAB", "."], "sentence-detokenized": "Bien qu'il soit principalement utilis\u00e9 par les statisticiens et autres praticiens n\u00e9cessitant un environnement pour le calcul statistique et le d\u00e9veloppement de logiciels, R peut \u00e9galement fonctionner comme une bo\u00eete \u00e0 outils g\u00e9n\u00e9rale pour le calcul matriciel - avec des performances comparables \u00e0 celles de GNU Octave ou MATLAB.", "token2charspan": [[0, 4], [5, 8], [8, 10], [11, 15], [16, 30], [31, 38], [39, 42], [43, 46], [47, 60], [61, 63], [64, 70], [71, 81], [82, 93], [94, 96], [97, 110], [111, 115], [116, 118], [119, 125], [126, 137], [138, 140], [141, 143], [144, 157], [158, 160], [161, 170], [170, 171], [172, 173], [174, 178], [179, 188], [189, 200], [201, 206], [207, 210], [211, 216], [217, 218], [219, 225], [226, 234], [235, 239], [240, 242], [243, 249], [250, 259], [260, 261], [262, 266], [267, 270], [271, 283], [284, 295], [296, 297], [298, 304], [305, 307], [308, 311], [312, 318], [319, 321], [322, 328], [328, 329]]}
{"doc_key": "ai-train-45", "ner": [[0, 1, "algorithm"], [6, 8, "field"], [12, 13, "misc"], [14, 15, "researcher"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[0, 1, 6, 8, "part-of", "", false, false], [0, 1, 14, 15, "origin", "", false, false], [12, 13, 14, 15, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["L'", "h\u00e9t\u00e9rodynage", "est", "une", "technique", "de", "traitement", "du", "signal", "invent\u00e9e", "par", "l'", "ing\u00e9nieur-inventeur", "canadien", "Reginald", "Fessenden", "qui", "permet", "de", "cr\u00e9er", "de", "nouvelles", "fr\u00e9quences", "en", "m\u00e9langeant", "deux", "fr\u00e9quences", "."], "sentence-detokenized": "L'h\u00e9t\u00e9rodynage est une technique de traitement du signal invent\u00e9e par l'ing\u00e9nieur-inventeur canadien Reginald Fessenden qui permet de cr\u00e9er de nouvelles fr\u00e9quences en m\u00e9langeant deux fr\u00e9quences.", "token2charspan": [[0, 2], [2, 14], [15, 18], [19, 22], [23, 32], [33, 35], [36, 46], [47, 49], [50, 56], [57, 65], [66, 69], [70, 72], [72, 91], [92, 100], [101, 109], [110, 119], [120, 123], [124, 130], [131, 133], [134, 139], [140, 142], [143, 152], [153, 163], [164, 166], [167, 177], [178, 182], [183, 193], [193, 194]]}
{"doc_key": "ai-train-46", "ner": [[21, 22, "person"], [16, 18, "misc"], [26, 27, "organisation"], [34, 34, "organisation"], [30, 32, "misc"], [36, 37, "person"], [44, 44, "organisation"], [40, 42, "misc"], [46, 47, "person"], [49, 50, "person"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "relations": [[21, 22, 16, 18, "role", "actor_in", false, false], [16, 18, 26, 27, "artifact", "", false, false], [30, 32, 34, 34, "artifact", "", false, false], [36, 37, 30, 32, "role", "actor_in", false, false], [40, 42, 44, 44, "artifact", "", false, false], [46, 47, 40, 42, "role", "actor_in", false, false], [49, 50, 40, 42, "role", "actor_in", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "sentence": ["Plusieurs", "autres", "films", "ont", "contribu\u00e9", "\u00e0", "remettre", "la", "3D", "sur", "la", "carte", "ce", "mois", "-l\u00e0", ":", "Hondo", ",", "un", "film", "de", "John", "Wayne", "(", "distribu\u00e9", "par", "Warner", "Bros.", ")", ",", "Miss", "Sadie", "Thompson", "de", "Columbia", "avec", "Rita", "Hayworth", ",", "et", "Money", "From", "Home", "de", "Paramount", "avec", "Dean", "Martin", "et", "Jerry", "Lewis", "."], "sentence-detokenized": "Plusieurs autres films ont contribu\u00e9 \u00e0 remettre la 3D sur la carte ce mois-l\u00e0 : Hondo, un film de John Wayne (distribu\u00e9 par Warner Bros.), Miss Sadie Thompson de Columbia avec Rita Hayworth, et Money From Home de Paramount avec Dean Martin et Jerry Lewis.", "token2charspan": [[0, 9], [10, 16], [17, 22], [23, 26], [27, 36], [37, 38], [39, 47], [48, 50], [51, 53], [54, 57], [58, 60], [61, 66], [67, 69], [70, 74], [74, 77], [78, 79], [80, 85], [85, 86], [87, 89], [90, 94], [95, 97], [98, 102], [103, 108], [109, 110], [110, 119], [120, 123], [124, 130], [131, 136], [136, 137], [137, 138], [139, 143], [144, 149], [150, 158], [159, 161], [162, 170], [171, 175], [176, 180], [181, 189], [189, 190], [191, 193], [194, 199], [200, 204], [205, 209], [210, 212], [213, 222], [223, 227], [228, 232], [233, 239], [240, 242], [243, 248], [249, 254], [254, 255]]}
{"doc_key": "ai-train-47", "ner": [[0, 0, "product"], [8, 9, "field"], [5, 7, "task"], [17, 17, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[0, 0, 5, 7, "general-affiliation", "", false, false], [0, 0, 17, 17, "artifact", "", false, false], [5, 7, 8, 9, "part-of", "task_part_of_field", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["DeepFace", "est", "un", "syst\u00e8me", "de", "reconnaissance", "faciale", "par", "apprentissage", "profond", "cr\u00e9\u00e9", "par", "un", "groupe", "de", "recherche", "de", "Facebook", "."], "sentence-detokenized": "DeepFace est un syst\u00e8me de reconnaissance faciale par apprentissage profond cr\u00e9\u00e9 par un groupe de recherche de Facebook.", "token2charspan": [[0, 8], [9, 12], [13, 15], [16, 23], [24, 26], [27, 41], [42, 49], [50, 53], [54, 67], [68, 75], [76, 80], [81, 84], [85, 87], [88, 94], [95, 97], [98, 107], [108, 110], [111, 119], [119, 120]]}
{"doc_key": "ai-train-48", "ner": [[1, 4, "field"], [12, 12, "conference"], [20, 20, "field"], [27, 34, "conference"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[1, 4, 20, 20, "part-of", "subfield", false, false], [12, 12, 1, 4, "topic", "", false, false], [27, 34, 1, 4, "topic", "", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["Le", "traitement", "de", "la", "g\u00e9om\u00e9trie", "est", "un", "sujet", "de", "recherche", "courant", "au", "SIGGRAPH", ",", "la", "principale", "conf\u00e9rence", "universitaire", "sur", "l'", "infographie", ",", "et", "le", "sujet", "principal", "du", "Symposium", "annuel", "sur", "le", "traitement", "de", "la", "g\u00e9om\u00e9trie", "."], "sentence-detokenized": "Le traitement de la g\u00e9om\u00e9trie est un sujet de recherche courant au SIGGRAPH, la principale conf\u00e9rence universitaire sur l'infographie, et le sujet principal du Symposium annuel sur le traitement de la g\u00e9om\u00e9trie.", "token2charspan": [[0, 2], [3, 13], [14, 16], [17, 19], [20, 29], [30, 33], [34, 36], [37, 42], [43, 45], [46, 55], [56, 63], [64, 66], [67, 75], [75, 76], [77, 79], [80, 90], [91, 101], [102, 115], [116, 119], [120, 122], [122, 133], [133, 134], [135, 137], [138, 140], [141, 146], [147, 156], [157, 159], [160, 169], [170, 176], [177, 180], [181, 183], [184, 194], [195, 197], [198, 200], [201, 210], [210, 211]]}
{"doc_key": "ai-train-49", "ner": [[0, 3, "task"], [6, 8, "task"], [21, 24, "algorithm"], [26, 26, "algorithm"], [30, 32, "algorithm"], [34, 34, "algorithm"], [38, 41, "algorithm"], [43, 43, "algorithm"], [48, 48, "misc"], [55, 57, "algorithm"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "relations": [[21, 24, 48, 48, "general-affiliation", "", false, false], [26, 26, 21, 24, "named", "", false, false], [30, 32, 48, 48, "general-affiliation", "", false, false], [34, 34, 30, 32, "named", "", false, false], [38, 41, 48, 48, "general-affiliation", "", false, false], [43, 43, 38, 41, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5], "sentence": ["L'", "extraction", "de", "caract\u00e9ristiques", "et", "la", "r\u00e9duction", "de", "dimension", "peuvent", "\u00eatre", "combin\u00e9es", "en", "une", "seule", "\u00e9tape", "en", "utilisant", "les", "techniques", "d'", "analyse", "en", "composantes", "principales", "(", "ACP", ")", ",", "d'", "analyse", "discriminante", "lin\u00e9aire", "(", "ADL", ")", "ou", "d'", "analyse", "de", "corr\u00e9lation", "canonique", "(", "ACC", ")", "comme", "\u00e9tape", "de", "pr\u00e9traitement", ",", "suivie", "d'", "un", "regroupement", "par", "k", "-", "NN", "sur", "les", "vecteurs", "de", "caract\u00e9ristiques", "dans", "un", "espace", "de", "dimension", "r\u00e9duite", "."], "sentence-detokenized": "L'extraction de caract\u00e9ristiques et la r\u00e9duction de dimension peuvent \u00eatre combin\u00e9es en une seule \u00e9tape en utilisant les techniques d'analyse en composantes principales (ACP), d'analyse discriminante lin\u00e9aire (ADL) ou d'analyse de corr\u00e9lation canonique (ACC) comme \u00e9tape de pr\u00e9traitement, suivie d'un regroupement par k -NN sur les vecteurs de caract\u00e9ristiques dans un espace de dimension r\u00e9duite.", "token2charspan": [[0, 2], [2, 12], [13, 15], [16, 32], [33, 35], [36, 38], [39, 48], [49, 51], [52, 61], [62, 69], [70, 74], [75, 84], [85, 87], [88, 91], [92, 97], [98, 103], [104, 106], [107, 116], [117, 120], [121, 131], [132, 134], [134, 141], [142, 144], [145, 156], [157, 168], [169, 170], [170, 173], [173, 174], [174, 175], [176, 178], [178, 185], [186, 199], [200, 208], [209, 210], [210, 213], [213, 214], [215, 217], [218, 220], [220, 227], [228, 230], [231, 242], [243, 252], [253, 254], [254, 257], [257, 258], [259, 264], [265, 270], [271, 273], [274, 287], [287, 288], [289, 295], [296, 298], [298, 300], [301, 313], [314, 317], [318, 319], [320, 321], [321, 323], [324, 327], [328, 331], [332, 340], [341, 343], [344, 360], [361, 365], [366, 368], [369, 375], [376, 378], [379, 388], [389, 396], [396, 397]]}
{"doc_key": "ai-train-50", "ner": [[0, 3, "algorithm"], [12, 13, "field"], [16, 18, "field"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[0, 3, 12, 13, "related-to", "good_at", true, false], [0, 3, 16, 18, "related-to", "good_at", true, false]], "relations_mapping_to_source": [0, 1], "sentence": ["Les", "r\u00e9seaux", "neuronaux", "artificiels", "sont", "des", "mod\u00e8les", "informatiques", "qui", "excellent", "dans", "l'", "apprentissage", "automatique", "et", "la", "reconnaissance", "des", "formes", "."], "sentence-detokenized": "Les r\u00e9seaux neuronaux artificiels sont des mod\u00e8les informatiques qui excellent dans l'apprentissage automatique et la reconnaissance des formes.", "token2charspan": [[0, 3], [4, 11], [12, 21], [22, 33], [34, 38], [39, 42], [43, 50], [51, 64], [65, 68], [69, 78], [79, 83], [84, 86], [86, 99], [100, 111], [112, 114], [115, 117], [118, 132], [133, 136], [137, 143], [143, 144]]}
{"doc_key": "ai-train-51", "ner": [[0, 1, "researcher"], [3, 4, "researcher"], [6, 10, "misc"], [12, 16, "conference"], [18, 18, "conference"], [35, 38, "algorithm"], [39, 40, "researcher"], [42, 43, "researcher"], [45, 51, "misc"], [53, 62, "conference"], [64, 64, "conference"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10], "relations": [[6, 10, 0, 1, "artifact", "", false, false], [6, 10, 3, 4, "artifact", "", false, false], [6, 10, 12, 16, "temporal", "", false, false], [18, 18, 12, 16, "named", "", false, false], [45, 51, 35, 38, "topic", "", false, false], [45, 51, 39, 40, "artifact", "", false, false], [45, 51, 42, 43, "artifact", "", false, false], [45, 51, 53, 62, "temporal", "", false, false], [64, 64, 53, 62, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "sentence": ["C.", "Papageorgiou", "et", "T.", "Poggio", ",", "A", "Trainable", "Pedestrian", "Detection", "system", ",", "International", "Journal", "of", "Computer", "Vision", "(", "IJCV", ")", ",", "pages", "1", ":", "15-33", ",", "2000", "d'", "autres", "utilisent", "des", "caract\u00e9ristiques", "locales", "comme", "l'", "histogramme", "des", "gradients", "orient\u00e9s", "N.", "Dalal", ",", "B.", "Triggs", ",", "Histograms", "of", "oriented", "gradients", "for", "human", "detection", ",", "IEEE", "Computer", "Society", "Conference", "on", "Computer", "Vision", "and", "Pattern", "Recognition", "(", "CVPR", ")", ",", "pages", "1", ":", "886-893", ",", "2005", "descripteurs", "."], "sentence-detokenized": "C. Papageorgiou et T. Poggio, A Trainable Pedestrian Detection system, International Journal of Computer Vision (IJCV), pages 1 : 15-33, 2000 d'autres utilisent des caract\u00e9ristiques locales comme l'histogramme des gradients orient\u00e9s N. Dalal, B. Triggs, Histograms of oriented gradients for human detection, IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR), pages 1 : 886-893, 2005 descripteurs.", "token2charspan": [[0, 2], [3, 15], [16, 18], [19, 21], [22, 28], [28, 29], [30, 31], [32, 41], [42, 52], [53, 62], [63, 69], [69, 70], [71, 84], [85, 92], [93, 95], [96, 104], [105, 111], [112, 113], [113, 117], [117, 118], [118, 119], [120, 125], [126, 127], [128, 129], [130, 135], [135, 136], [137, 141], [142, 144], [144, 150], [151, 160], [161, 164], [165, 181], [182, 189], [190, 195], [196, 198], [198, 209], [210, 213], [214, 223], [224, 232], [233, 235], [236, 241], [241, 242], [243, 245], [246, 252], [252, 253], [254, 264], [265, 267], [268, 276], [277, 286], [287, 290], [291, 296], [297, 306], [306, 307], [308, 312], [313, 321], [322, 329], [330, 340], [341, 343], [344, 352], [353, 359], [360, 363], [364, 371], [372, 383], [384, 385], [385, 389], [389, 390], [390, 391], [392, 397], [398, 399], [400, 401], [402, 409], [409, 410], [411, 415], [416, 428], [428, 429]]}
{"doc_key": "ai-train-52", "ner": [[1, 1, "algorithm"], [6, 8, "algorithm"], [13, 15, "task"], [18, 19, "field"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[1, 1, 6, 8, "type-of", "", false, false], [13, 15, 1, 1, "usage", "", true, false], [13, 15, 18, 19, "part-of", "task_part_of_field", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["Un", "autoencodeur", "est", "un", "type", "de", "r\u00e9seau", "neuronal", "artificiel", "utilis\u00e9", "pour", "apprendre", "l'", "apprentissage", "des", "fonctionnalit\u00e9s", "de", "mani\u00e8re", "non", "supervis\u00e9e", "."], "sentence-detokenized": "Un autoencodeur est un type de r\u00e9seau neuronal artificiel utilis\u00e9 pour apprendre l'apprentissage des fonctionnalit\u00e9s de mani\u00e8re non supervis\u00e9e.", "token2charspan": [[0, 2], [3, 15], [16, 19], [20, 22], [23, 27], [28, 30], [31, 37], [38, 46], [47, 57], [58, 65], [66, 70], [71, 80], [81, 83], [83, 96], [97, 100], [101, 116], [117, 119], [120, 127], [128, 131], [132, 142], [142, 143]]}
{"doc_key": "ai-train-53", "ner": [[0, 1, "researcher"], [6, 6, "organisation"], [15, 17, "field"], [20, 22, "field"], [27, 31, "organisation"], [33, 33, "organisation"], [43, 45, "field"], [48, 50, "field"], [58, 58, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "relations": [[0, 1, 6, 6, "role", "fellow_of", false, false], [0, 1, 15, 17, "related-to", "contributes_to", false, false], [0, 1, 20, 22, "related-to", "contributes_to", false, false], [0, 1, 27, 31, "role", "fellow_of", false, false], [0, 1, 43, 45, "related-to", "contributes_to", false, false], [0, 1, 48, 50, "related-to", "contributes_to", false, false], [33, 33, 27, 31, "named", "", false, false], [58, 58, 27, 31, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "sentence": ["M.", "Haralick", "est", "membre", "de", "l'", "IEEE", "pour", "ses", "contributions", "dans", "le", "domaine", "de", "la", "vision", "par", "ordinateur", "et", "du", "traitement", "des", "images", "et", "membre", "de", "l'", "International", "Association", "for", "Pattern", "Recognition", "(", "IAPR", ")", "pour", "ses", "contributions", "dans", "le", "domaine", "de", "la", "reconnaissance", "des", "formes", ",", "du", "traitement", "des", "images", "et", "pour", "les", "services", "rendus", "\u00e0", "l'", "IAPR."], "sentence-detokenized": "M. Haralick est membre de l'IEEE pour ses contributions dans le domaine de la vision par ordinateur et du traitement des images et membre de l'International Association for Pattern Recognition (IAPR) pour ses contributions dans le domaine de la reconnaissance des formes, du traitement des images et pour les services rendus \u00e0 l'IAPR.", "token2charspan": [[0, 2], [3, 11], [12, 15], [16, 22], [23, 25], [26, 28], [28, 32], [33, 37], [38, 41], [42, 55], [56, 60], [61, 63], [64, 71], [72, 74], [75, 77], [78, 84], [85, 88], [89, 99], [100, 102], [103, 105], [106, 116], [117, 120], [121, 127], [128, 130], [131, 137], [138, 140], [141, 143], [143, 156], [157, 168], [169, 172], [173, 180], [181, 192], [193, 194], [194, 198], [198, 199], [200, 204], [205, 208], [209, 222], [223, 227], [228, 230], [231, 238], [239, 241], [242, 244], [245, 259], [260, 263], [264, 270], [270, 271], [272, 274], [275, 285], [286, 289], [290, 296], [297, 299], [300, 304], [305, 308], [309, 317], [318, 324], [325, 326], [327, 329], [329, 334]]}
{"doc_key": "ai-train-54", "ner": [[4, 8, "task"], [18, 20, "algorithm"], [22, 22, "algorithm"], [26, 27, "researcher"], [29, 30, "organisation"], [32, 33, "researcher"], [36, 38, "university"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "relations": [[4, 8, 18, 20, "usage", "", false, false], [18, 20, 26, 27, "origin", "", true, false], [18, 20, 32, 33, "origin", "", true, false], [22, 22, 18, 20, "named", "", false, false], [26, 27, 29, 30, "physical", "", false, false], [26, 27, 29, 30, "role", "", false, false], [32, 33, 36, 38, "physical", "", false, false], [32, 33, 36, 38, "role", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "sentence": ["La", "premi\u00e8re", "tentative", "de", "RAS", "de", "bout", "en", "bout", "a", "\u00e9t\u00e9", "r\u00e9alis\u00e9e", "avec", "des", "syst\u00e8mes", "bas\u00e9s", "sur", "la", "classification", "temporelle", "connexionniste", "(", "CTC", ")", "pr\u00e9sent\u00e9s", "par", "Alex", "Graves", "de", "Google", "DeepMind", "et", "Navdeep", "Jaitly", "de", "l'", "Universit\u00e9", "de", "Toronto", "en", "2014", "."], "sentence-detokenized": "La premi\u00e8re tentative de RAS de bout en bout a \u00e9t\u00e9 r\u00e9alis\u00e9e avec des syst\u00e8mes bas\u00e9s sur la classification temporelle connexionniste (CTC) pr\u00e9sent\u00e9s par Alex Graves de Google DeepMind et Navdeep Jaitly de l'Universit\u00e9 de Toronto en 2014.", "token2charspan": [[0, 2], [3, 11], [12, 21], [22, 24], [25, 28], [29, 31], [32, 36], [37, 39], [40, 44], [45, 46], [47, 50], [51, 59], [60, 64], [65, 68], [69, 77], [78, 83], [84, 87], [88, 90], [91, 105], [106, 116], [117, 131], [132, 133], [133, 136], [136, 137], [138, 147], [148, 151], [152, 156], [157, 163], [164, 166], [167, 173], [174, 182], [183, 185], [186, 193], [194, 200], [201, 203], [204, 206], [206, 216], [217, 219], [220, 227], [228, 230], [231, 235], [235, 236]]}
{"doc_key": "ai-train-55", "ner": [[0, 2, "algorithm"], [4, 4, "algorithm"], [11, 12, "algorithm"], [14, 14, "algorithm"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[4, 4, 0, 2, "named", "", false, false], [11, 12, 0, 2, "type-of", "", false, false], [14, 14, 11, 12, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["La", "programmation", "lin\u00e9aire-fractionnelle", "(", "LFP", ")", "est", "une", "g\u00e9n\u00e9ralisation", "de", "la", "programmation", "lin\u00e9aire", "(", "LP", ")", "."], "sentence-detokenized": "La programmation lin\u00e9aire-fractionnelle (LFP) est une g\u00e9n\u00e9ralisation de la programmation lin\u00e9aire (LP).", "token2charspan": [[0, 2], [3, 16], [17, 39], [40, 41], [41, 44], [44, 45], [46, 49], [50, 53], [54, 68], [69, 71], [72, 74], [75, 88], [89, 97], [98, 99], [99, 101], [101, 102], [102, 103]]}
{"doc_key": "ai-train-56", "ner": [[0, 1, "researcher"], [10, 11, "misc"], [14, 21, "conference"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[0, 1, 10, 11, "win-defeat", "", false, false], [10, 11, 14, 21, "temporal", "", false, false]], "relations_mapping_to_source": [0, 1], "sentence": ["M.", "Lafferty", "a", "re\u00e7u", "de", "nombreux", "prix", ",", "dont", "deux", "prix", "Test-of-Time", "\u00e0", "l'", "International", "Conference", "on", "Machine", "Learning", "2011", "&", "2012", ","], "sentence-detokenized": "M. Lafferty a re\u00e7u de nombreux prix, dont deux prix Test-of-Time \u00e0 l'International Conference on Machine Learning 2011 & 2012,", "token2charspan": [[0, 2], [3, 11], [12, 13], [14, 18], [19, 21], [22, 30], [31, 35], [35, 36], [37, 41], [42, 46], [47, 51], [52, 64], [65, 66], [67, 69], [69, 82], [83, 93], [94, 96], [97, 104], [105, 113], [114, 118], [119, 120], [121, 125], [125, 126]]}
{"doc_key": "ai-train-57", "ner": [[11, 11, "product"], [13, 13, "programlang"], [28, 29, "algorithm"]], "ner_mapping_to_source": [0, 1, 2], "relations": [], "relations_mapping_to_source": [], "sentence": ["Avec", "l'", "av\u00e8nement", "des", "cadres", "bas\u00e9s", "sur", "des", "composants", "tels", "que", ".NET", "et", "Java", ",", "les", "environnements", "de", "d\u00e9veloppement", "bas\u00e9s", "sur", "des", "composants", "sont", "capables", "de", "d\u00e9ployer", "le", "r\u00e9seau", "neuronal", "d\u00e9velopp\u00e9", "dans", "ces", "cadres", "en", "tant", "que", "composants", "h\u00e9ritables", "."], "sentence-detokenized": "Avec l'av\u00e8nement des cadres bas\u00e9s sur des composants tels que .NET et Java, les environnements de d\u00e9veloppement bas\u00e9s sur des composants sont capables de d\u00e9ployer le r\u00e9seau neuronal d\u00e9velopp\u00e9 dans ces cadres en tant que composants h\u00e9ritables.", "token2charspan": [[0, 4], [5, 7], [7, 16], [17, 20], [21, 27], [28, 33], [34, 37], [38, 41], [42, 52], [53, 57], [58, 61], [62, 66], [67, 69], [70, 74], [74, 75], [76, 79], [80, 94], [95, 97], [98, 111], [112, 117], [118, 121], [122, 125], [126, 136], [137, 141], [142, 150], [151, 153], [154, 162], [163, 165], [166, 172], [173, 181], [182, 191], [192, 196], [197, 200], [201, 207], [208, 210], [211, 215], [216, 219], [220, 230], [231, 241], [241, 242]]}
{"doc_key": "ai-train-58", "ner": [[3, 3, "metrics"]], "ner_mapping_to_source": [0], "relations": [], "relations_mapping_to_source": [], "sentence": ["Comme", "pour", "le", "BLEU", ",", "l'", "unit\u00e9", "de", "base", "de", "l'", "\u00e9valuation", "est", "la", "phrase", ".", "L'", "algorithme", "cr\u00e9e", "d'", "abord", "un", "alignement", "(", "voir", "illustrations", ")", "entre", "deux", "phrases", ",", "la", "cha\u00eene", "de", "traduction", "candidate", "et", "la", "cha\u00eene", "de", "traduction", "de", "r\u00e9f\u00e9rence", "."], "sentence-detokenized": "Comme pour le BLEU, l'unit\u00e9 de base de l'\u00e9valuation est la phrase. L'algorithme cr\u00e9e d'abord un alignement (voir illustrations) entre deux phrases, la cha\u00eene de traduction candidate et la cha\u00eene de traduction de r\u00e9f\u00e9rence.", "token2charspan": [[0, 5], [6, 10], [11, 13], [14, 18], [18, 19], [20, 22], [22, 27], [28, 30], [31, 35], [36, 38], [39, 41], [41, 51], [52, 55], [56, 58], [59, 65], [65, 66], [67, 69], [69, 79], [80, 84], [85, 87], [87, 92], [93, 95], [96, 106], [107, 108], [108, 112], [113, 126], [126, 127], [128, 133], [134, 138], [139, 146], [146, 147], [148, 150], [151, 157], [158, 160], [161, 171], [172, 181], [182, 184], [185, 187], [188, 194], [195, 197], [198, 208], [209, 211], [212, 221], [221, 222]]}
{"doc_key": "ai-train-59", "ner": [[6, 15, "conference"], [30, 30, "task"], [28, 33, "task"], [37, 38, "metrics"], [40, 44, "metrics"], [49, 52, "conference"], [54, 54, "conference"], [57, 57, "location"], [59, 59, "country"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "relations": [[6, 15, 30, 30, "related-to", "subject_at", false, false], [6, 15, 28, 33, "related-to", "subject_at", false, false], [37, 38, 6, 15, "temporal", "", false, false], [40, 44, 37, 38, "named", "", true, false], [54, 54, 49, 52, "named", "", false, false], [57, 57, 59, 59, "physical", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5], "sentence": ["L'", "une", "des", "m\u00e9triques", "utilis\u00e9es", "lors", "des", "conf\u00e9rences", "annuelles", "du", "NIST", "sur", "la", "compr\u00e9hension", "des", "documents", ",", "dans", "lesquelles", "des", "groupes", "de", "recherche", "soumettent", "leurs", "syst\u00e8mes", "pour", "des", "t\u00e2ches", "de", "r\u00e9sum\u00e9", "et", "de", "traduction", ",", "est", "la", "m\u00e9trique", "ROUGE", "(", "Recall-Oriented", "Understudy", "for", "Gisting", "Evaluation", ",", "In", "Advances", "of", "Neural", "Information", "Processing", "Systems", "(", "NIPS", ")", ",", "Montr\u00e9al", ",", "Canada", ",", "d\u00e9cembre", "-", "2014", ")", "."], "sentence-detokenized": "L'une des m\u00e9triques utilis\u00e9es lors des conf\u00e9rences annuelles du NIST sur la compr\u00e9hension des documents, dans lesquelles des groupes de recherche soumettent leurs syst\u00e8mes pour des t\u00e2ches de r\u00e9sum\u00e9 et de traduction, est la m\u00e9trique ROUGE (Recall-Oriented Understudy for Gisting Evaluation, In Advances of Neural Information Processing Systems (NIPS), Montr\u00e9al, Canada, d\u00e9cembre - 2014).", "token2charspan": [[0, 2], [2, 5], [6, 9], [10, 19], [20, 29], [30, 34], [35, 38], [39, 50], [51, 60], [61, 63], [64, 68], [69, 72], [73, 75], [76, 89], [90, 93], [94, 103], [103, 104], [105, 109], [110, 120], [121, 124], [125, 132], [133, 135], [136, 145], [146, 156], [157, 162], [163, 171], [172, 176], [177, 180], [181, 187], [188, 190], [191, 197], [198, 200], [201, 203], [204, 214], [214, 215], [216, 219], [220, 222], [223, 231], [232, 237], [238, 239], [239, 254], [255, 265], [266, 269], [270, 277], [278, 288], [288, 289], [290, 292], [293, 301], [302, 304], [305, 311], [312, 323], [324, 334], [335, 342], [343, 344], [344, 348], [348, 349], [349, 350], [351, 359], [359, 360], [361, 367], [367, 368], [369, 377], [378, 379], [380, 384], [384, 385], [385, 386]]}
{"doc_key": "ai-train-60", "ner": [[6, 6, "programlang"], [8, 8, "product"], [10, 11, "programlang"], [15, 15, "product"], [21, 21, "programlang"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[6, 6, 10, 11, "type-of", "", false, false], [6, 6, 21, 21, "named", "", false, false], [8, 8, 10, 11, "part-of", "", false, false], [8, 8, 15, 15, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["M\u00eame", "impl\u00e9mentation", ",", "\u00e0", "ex\u00e9cuter", "en", "Java", "avec", "JShell", "(", "Java", "9", "minimum", ")", ":", "codejshell", "scriptfile", "/", "codesyntaxhighlight", "lang", "=", "java"], "sentence-detokenized": "M\u00eame impl\u00e9mentation, \u00e0 ex\u00e9cuter en Java avec JShell (Java 9 minimum) : codejshell scriptfile / codesyntaxhighlight lang = java", "token2charspan": [[0, 4], [5, 19], [19, 20], [21, 22], [23, 31], [32, 34], [35, 39], [40, 44], [45, 51], [52, 53], [53, 57], [58, 59], [60, 67], [67, 68], [69, 70], [71, 81], [82, 92], [93, 94], [95, 114], [115, 119], [120, 121], [122, 126]]}
{"doc_key": "ai-train-61", "ner": [[0, 3, "metrics"], [8, 9, "metrics"]], "ner_mapping_to_source": [0, 1], "relations": [[0, 3, 8, 9, "origin", "based_on", false, false]], "relations_mapping_to_source": [0], "sentence": ["La", "m\u00e9trique", "du", "NIST", "est", "bas\u00e9e", "sur", "la", "m\u00e9trique", "BLEU", ",", "mais", "avec", "quelques", "modifications", "."], "sentence-detokenized": "La m\u00e9trique du NIST est bas\u00e9e sur la m\u00e9trique BLEU, mais avec quelques modifications.", "token2charspan": [[0, 2], [3, 11], [12, 14], [15, 19], [20, 23], [24, 29], [30, 33], [34, 36], [37, 45], [46, 50], [50, 51], [52, 56], [57, 61], [62, 70], [71, 84], [84, 85]]}
{"doc_key": "ai-train-62", "ner": [[9, 9, "country"], [12, 14, "university"], [17, 19, "university"], [27, 28, "product"], [33, 34, "algorithm"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[12, 14, 9, 9, "physical", "", false, false], [17, 19, 9, 9, "physical", "", false, false], [27, 28, 12, 14, "origin", "", false, false], [27, 28, 17, 19, "origin", "", false, false], [27, 28, 33, 34, "type-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4], "sentence": ["\u00c0", "la", "fin", "des", "ann\u00e9es", "1980", ",", "deux", "universit\u00e9s", "n\u00e9erlandaises", ",", "l'", "Universit\u00e9", "de", "Groningue", "et", "l'", "Universit\u00e9", "de", "Twente", ",", "ont", "lanc\u00e9", "conjointement", "un", "projet", "appel\u00e9", "Knowledge", "Graphs", ",", "qui", "sont", "des", "r\u00e9seaux", "s\u00e9mantiques", ",", "mais", "avec", "la", "contrainte", "suppl\u00e9mentaire", "que", "les", "ar\u00eates", "sont", "limit\u00e9es", "\u00e0", "un", "ensemble", "restreint", "de", "relations", "possibles", ",", "afin", "de", "faciliter", "les", "alg\u00e8bres", "sur", "le", "graphe", "."], "sentence-detokenized": "\u00c0 la fin des ann\u00e9es 1980, deux universit\u00e9s n\u00e9erlandaises, l'Universit\u00e9 de Groningue et l'Universit\u00e9 de Twente, ont lanc\u00e9 conjointement un projet appel\u00e9 Knowledge Graphs, qui sont des r\u00e9seaux s\u00e9mantiques, mais avec la contrainte suppl\u00e9mentaire que les ar\u00eates sont limit\u00e9es \u00e0 un ensemble restreint de relations possibles, afin de faciliter les alg\u00e8bres sur le graphe.", "token2charspan": [[0, 1], [2, 4], [5, 8], [9, 12], [13, 19], [20, 24], [24, 25], [26, 30], [31, 42], [43, 56], [56, 57], [58, 60], [60, 70], [71, 73], [74, 83], [84, 86], [87, 89], [89, 99], [100, 102], [103, 109], [109, 110], [111, 114], [115, 120], [121, 134], [135, 137], [138, 144], [145, 151], [152, 161], [162, 168], [168, 169], [170, 173], [174, 178], [179, 182], [183, 190], [191, 202], [202, 203], [204, 208], [209, 213], [214, 216], [217, 227], [228, 242], [243, 246], [247, 250], [251, 257], [258, 262], [263, 271], [272, 273], [274, 276], [277, 285], [286, 295], [296, 298], [299, 308], [309, 318], [318, 319], [320, 324], [325, 327], [328, 337], [338, 341], [342, 350], [351, 354], [355, 357], [358, 364], [364, 365]]}
{"doc_key": "ai-train-63", "ner": [[1, 2, "product"], [17, 19, "product"]], "ner_mapping_to_source": [0, 1], "relations": [[1, 2, 17, 19, "part-of", "", false, false]], "relations_mapping_to_source": [0], "sentence": ["Les", "correcteurs", "grammaticaux", "sont", "le", "plus", "souvent", "int\u00e9gr\u00e9s", "\u00e0", "un", "programme", "plus", "vaste", ",", "tel", "qu'", "un", "traitement", "de", "texte", ",", "mais", "ils", "sont", "\u00e9galement", "disponibles", "sous", "la", "forme", "d'", "une", "application", "autonome", "qui", "peut", "\u00eatre", "activ\u00e9e", "\u00e0", "partir", "de", "programmes", "qui", "traitent", "du", "texte", "\u00e9ditable", "."], "sentence-detokenized": "Les correcteurs grammaticaux sont le plus souvent int\u00e9gr\u00e9s \u00e0 un programme plus vaste, tel qu'un traitement de texte, mais ils sont \u00e9galement disponibles sous la forme d'une application autonome qui peut \u00eatre activ\u00e9e \u00e0 partir de programmes qui traitent du texte \u00e9ditable.", "token2charspan": [[0, 3], [4, 15], [16, 28], [29, 33], [34, 36], [37, 41], [42, 49], [50, 58], [59, 60], [61, 63], [64, 73], [74, 78], [79, 84], [84, 85], [86, 89], [90, 93], [93, 95], [96, 106], [107, 109], [110, 115], [115, 116], [117, 121], [122, 125], [126, 130], [131, 140], [141, 152], [153, 157], [158, 160], [161, 166], [167, 169], [169, 172], [173, 184], [185, 193], [194, 197], [198, 202], [203, 207], [208, 215], [216, 217], [218, 224], [225, 227], [228, 238], [239, 242], [243, 251], [252, 254], [255, 260], [261, 269], [269, 270]]}
{"doc_key": "ai-train-64", "ner": [[5, 11, "organisation"], [15, 20, "conference"], [24, 26, "organisation"], [33, 35, "conference"], [37, 39, "conference"], [41, 43, "conference"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5], "relations": [], "relations_mapping_to_source": [], "sentence": ["Il", "est", "membre", "de", "l'", "American", "Association", "for", "the", "Advancement", "of", "Science", ",", "de", "l'", "Association", "for", "the", "Advancement", "Artificial", "Intelligence", "et", "de", "la", "Cognitive", "Science", "Society", ",", "et", "r\u00e9dacteur", "en", "chef", "de", "J.", "Automated", "Reasoning", ",", "J.", "Learning", "Sciences", "et", "J.", "Applied", "Ontology", "."], "sentence-detokenized": "Il est membre de l'American Association for the Advancement of Science, de l'Association for the Advancement Artificial Intelligence et de la Cognitive Science Society, et r\u00e9dacteur en chef de J. Automated Reasoning, J. Learning Sciences et J. Applied Ontology.", "token2charspan": [[0, 2], [3, 6], [7, 13], [14, 16], [17, 19], [19, 27], [28, 39], [40, 43], [44, 47], [48, 59], [60, 62], [63, 70], [70, 71], [72, 74], [75, 77], [77, 88], [89, 92], [93, 96], [97, 108], [109, 119], [120, 132], [133, 135], [136, 138], [139, 141], [142, 151], [152, 159], [160, 167], [167, 168], [169, 171], [172, 181], [182, 184], [185, 189], [190, 192], [193, 195], [196, 205], [206, 215], [215, 216], [217, 219], [220, 228], [229, 237], [238, 240], [241, 243], [244, 251], [252, 260], [260, 261]]}
{"doc_key": "ai-train-65", "ner": [[0, 3, "algorithm"], [5, 5, "algorithm"], [11, 14, "task"], [25, 26, "researcher"], [29, 31, "university"], [34, 35, "researcher"], [37, 40, "organisation"], [42, 42, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "relations": [[0, 3, 11, 14, "type-of", "", false, false], [0, 3, 25, 26, "origin", "", false, false], [0, 3, 34, 35, "origin", "", false, false], [5, 5, 0, 3, "named", "", false, false], [25, 26, 29, 31, "physical", "", false, false], [25, 26, 29, 31, "role", "", false, false], [34, 35, 37, 40, "role", "", false, false], [42, 42, 37, 40, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "sentence": ["Le", "codage", "pr\u00e9dictif", "lin\u00e9aire", "(", "LPC", ")", ",", "une", "forme", "de", "codage", "de", "la", "parole", ",", "a", "commenc\u00e9", "\u00e0", "se", "d\u00e9velopper", "avec", "les", "travaux", "de", "Fumitada", "Itakura", "de", "l'", "universit\u00e9", "de", "Nagoya", "et", "de", "Shuzo", "Saito", "de", "Nippon", "Telegraph", "and", "Telephone", "(", "NTT", ")", "en", "1966", "."], "sentence-detokenized": "Le codage pr\u00e9dictif lin\u00e9aire (LPC), une forme de codage de la parole, a commenc\u00e9 \u00e0 se d\u00e9velopper avec les travaux de Fumitada Itakura de l'universit\u00e9 de Nagoya et de Shuzo Saito de Nippon Telegraph and Telephone (NTT) en 1966.", "token2charspan": [[0, 2], [3, 9], [10, 19], [20, 28], [29, 30], [30, 33], [33, 34], [34, 35], [36, 39], [40, 45], [46, 48], [49, 55], [56, 58], [59, 61], [62, 68], [68, 69], [70, 71], [72, 80], [81, 82], [83, 85], [86, 96], [97, 101], [102, 105], [106, 113], [114, 116], [117, 125], [126, 133], [134, 136], [137, 139], [139, 149], [150, 152], [153, 159], [160, 162], [163, 165], [166, 171], [172, 177], [178, 180], [181, 187], [188, 197], [198, 201], [202, 211], [212, 213], [213, 216], [216, 217], [218, 220], [221, 225], [225, 226]]}
{"doc_key": "ai-train-66", "ner": [[56, 58, "metrics"]], "ner_mapping_to_source": [0], "relations": [], "relations_mapping_to_source": [], "sentence": ["Si", "le", "signal", "est", "en", "outre", "ergodique", ",", "tous", "les", "chemins", "d'", "\u00e9chantillonnage", "pr\u00e9sentent", "la", "m\u00eame", "moyenne", "temporelle", "et", "donc", "mathR", "_", "x", "^", "{n", "/", "T", "_", "0}", "(", "\\", "tau", ")", "=", "\\", "widehat", "{", "R}", "_", "x", "^", "{n", "/", "T", "_", "0}", "(", "\\", "tau", ")", "/", "math", "au", "sens", "de", "l'", "erreur", "quadratique", "moyenne", "."], "sentence-detokenized": "Si le signal est en outre ergodique, tous les chemins d'\u00e9chantillonnage pr\u00e9sentent la m\u00eame moyenne temporelle et donc mathR _ x ^ {n / T _ 0} (\\ tau) =\\ widehat {R} _ x ^ {n / T _ 0} (\\ tau) / math au sens de l'erreur quadratique moyenne.", "token2charspan": [[0, 2], [3, 5], [6, 12], [13, 16], [17, 19], [20, 25], [26, 35], [35, 36], [37, 41], [42, 45], [46, 53], [54, 56], [56, 71], [72, 82], [83, 85], [86, 90], [91, 98], [99, 109], [110, 112], [113, 117], [118, 123], [124, 125], [126, 127], [128, 129], [130, 132], [133, 134], [135, 136], [137, 138], [139, 141], [142, 143], [143, 144], [145, 148], [148, 149], [150, 151], [151, 152], [153, 160], [161, 162], [162, 164], [165, 166], [167, 168], [169, 170], [171, 173], [174, 175], [176, 177], [178, 179], [180, 182], [183, 184], [184, 185], [186, 189], [189, 190], [191, 192], [193, 197], [198, 200], [201, 205], [206, 208], [209, 211], [211, 217], [218, 229], [230, 237], [237, 238]]}
{"doc_key": "ai-train-67", "ner": [[0, 3, "task"], [6, 9, "task"], [22, 25, "algorithm"], [27, 27, "algorithm"], [32, 34, "algorithm"], [36, 36, "algorithm"], [41, 44, "algorithm"], [46, 46, "algorithm"], [52, 57, "algorithm"], [59, 59, "algorithm"], [62, 64, "misc"], [70, 70, "algorithm"], [73, 75, "misc"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12], "relations": [[22, 25, 62, 64, "related-to", "", false, false], [27, 27, 22, 25, "named", "", false, false], [32, 34, 62, 64, "related-to", "", false, false], [36, 36, 32, 34, "named", "", false, false], [41, 44, 62, 64, "related-to", "", false, false], [46, 46, 41, 44, "named", "", false, false], [52, 57, 62, 64, "related-to", "", false, false], [59, 59, 52, 57, "named", "", false, false], [70, 70, 73, 75, "related-to", "", true, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "sentence": ["L'", "extraction", "de", "caract\u00e9ristiques", "et", "la", "r\u00e9duction", "de", "la", "dimension", "peuvent", "\u00eatre", "combin\u00e9es", "en", "une", "seule", "\u00e9tape", "\u00e0", "l'", "aide", "de", "l'", "analyse", "en", "composantes", "principales", "(", "ACP", ")", ",", "de", "l'", "analyse", "discriminante", "lin\u00e9aire", "(", "ADL", ")", ",", "de", "l'", "analyse", "de", "corr\u00e9lation", "canonique", "(", "ACC", ")", "ou", "des", "techniques", "de", "factorisation", "de", "la", "matrice", "non", "n\u00e9gative", "(", "NMF", ")", "comme", "\u00e9tape", "de", "pr\u00e9traitement", ",", "suivie", "du", "regroupement", "par", "K-NN", "sur", "les", "vecteurs", "de", "caract\u00e9ristiques", "dans", "l'", "espace", "\u00e0", "dimension", "r\u00e9duite", "."], "sentence-detokenized": "L'extraction de caract\u00e9ristiques et la r\u00e9duction de la dimension peuvent \u00eatre combin\u00e9es en une seule \u00e9tape \u00e0 l'aide de l'analyse en composantes principales (ACP), de l'analyse discriminante lin\u00e9aire (ADL), de l'analyse de corr\u00e9lation canonique (ACC) ou des techniques de factorisation de la matrice non n\u00e9gative (NMF) comme \u00e9tape de pr\u00e9traitement, suivie du regroupement par K-NN sur les vecteurs de caract\u00e9ristiques dans l'espace \u00e0 dimension r\u00e9duite.", "token2charspan": [[0, 2], [2, 12], [13, 15], [16, 32], [33, 35], [36, 38], [39, 48], [49, 51], [52, 54], [55, 64], [65, 72], [73, 77], [78, 87], [88, 90], [91, 94], [95, 100], [101, 106], [107, 108], [109, 111], [111, 115], [116, 118], [119, 121], [121, 128], [129, 131], [132, 143], [144, 155], [156, 157], [157, 160], [160, 161], [161, 162], [163, 165], [166, 168], [168, 175], [176, 189], [190, 198], [199, 200], [200, 203], [203, 204], [204, 205], [206, 208], [209, 211], [211, 218], [219, 221], [222, 233], [234, 243], [244, 245], [245, 248], [248, 249], [250, 252], [253, 256], [257, 267], [268, 270], [271, 284], [285, 287], [288, 290], [291, 298], [299, 302], [303, 311], [312, 313], [313, 316], [316, 317], [318, 323], [324, 329], [330, 332], [333, 346], [346, 347], [348, 354], [355, 357], [358, 370], [371, 374], [375, 379], [380, 383], [384, 387], [388, 396], [397, 399], [400, 416], [417, 421], [422, 424], [424, 430], [431, 432], [433, 442], [443, 450], [450, 451]]}
{"doc_key": "ai-train-68", "ner": [[4, 4, "programlang"], [6, 6, "programlang"], [8, 8, "programlang"], [10, 10, "programlang"], [16, 16, "product"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[16, 16, 4, 4, "related-to", "program_type_compatible_with", false, false], [16, 16, 6, 6, "related-to", "program_type_compatible_with", false, false], [16, 16, 8, 8, "related-to", "program_type_compatible_with", false, false], [16, 16, 10, 10, "related-to", "program_type_compatible_with", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Les", "biblioth\u00e8ques", "\u00e9crites", "en", "Perl", ",", "Java", ",", "ActiveX", "ou", ".NET", "peuvent", "\u00eatre", "appel\u00e9es", "directement", "depuis", "MATLAB", ","], "sentence-detokenized": "Les biblioth\u00e8ques \u00e9crites en Perl, Java, ActiveX ou .NET peuvent \u00eatre appel\u00e9es directement depuis MATLAB,", "token2charspan": [[0, 3], [4, 17], [18, 25], [26, 28], [29, 33], [33, 34], [35, 39], [39, 40], [41, 48], [49, 51], [52, 56], [57, 64], [65, 69], [70, 78], [79, 90], [91, 97], [98, 104], [104, 105]]}
{"doc_key": "ai-train-69", "ner": [[4, 10, "task"], [13, 16, "task"], [36, 38, "task"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[4, 10, 13, 16, "part-of", "", false, false]], "relations_mapping_to_source": [0], "sentence": ["La", "t\u00e2che", "consistant", "\u00e0", "reconna\u00eetre", "des", "entit\u00e9s", "nomm\u00e9es", "dans", "un", "texte", "est", "appel\u00e9e", "reconnaissance", "d'", "entit\u00e9s", "nomm\u00e9es", ",", "tandis", "que", "la", "t\u00e2che", "consistant", "\u00e0", "d\u00e9terminer", "l'", "identit\u00e9", "des", "entit\u00e9s", "nomm\u00e9es", "mentionn\u00e9es", "dans", "un", "texte", "est", "appel\u00e9e", "liaison", "d'", "entit\u00e9s", "."], "sentence-detokenized": "La t\u00e2che consistant \u00e0 reconna\u00eetre des entit\u00e9s nomm\u00e9es dans un texte est appel\u00e9e reconnaissance d'entit\u00e9s nomm\u00e9es, tandis que la t\u00e2che consistant \u00e0 d\u00e9terminer l'identit\u00e9 des entit\u00e9s nomm\u00e9es mentionn\u00e9es dans un texte est appel\u00e9e liaison d'entit\u00e9s.", "token2charspan": [[0, 2], [3, 8], [9, 19], [20, 21], [22, 33], [34, 37], [38, 45], [46, 53], [54, 58], [59, 61], [62, 67], [68, 71], [72, 79], [80, 94], [95, 97], [97, 104], [105, 112], [112, 113], [114, 120], [121, 124], [125, 127], [128, 133], [134, 144], [145, 146], [147, 157], [158, 160], [160, 168], [169, 172], [173, 180], [181, 188], [189, 200], [201, 205], [206, 208], [209, 214], [215, 218], [219, 226], [227, 234], [235, 237], [237, 244], [244, 245]]}
{"doc_key": "ai-train-70", "ner": [[1, 2, "algorithm"], [33, 33, "programlang"], [35, 35, "algorithm"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[1, 2, 35, 35, "part-of", "", true, false], [35, 35, 33, 33, "part-of", "", true, false]], "relations_mapping_to_source": [0, 1], "sentence": ["Les", "fonctions", "sigmo\u00efdes", "et", "les", "d\u00e9riv\u00e9s", "utilis\u00e9s", "dans", "le", "paquetage", "\u00e9taient", "\u00e0", "l'", "origine", "inclus", "dans", "le", "paquetage", ",", "\u00e0", "partir", "de", "la", "version", "0.8.0", ",", "ils", "ont", "\u00e9t\u00e9", "publi\u00e9s", "dans", "un", "paquetage", "R", "s\u00e9par\u00e9", "sigmo\u00efde", ",", "dans", "le", "but", "de", "permettre", "une", "utilisation", "plus", "g\u00e9n\u00e9rale", "."], "sentence-detokenized": "Les fonctions sigmo\u00efdes et les d\u00e9riv\u00e9s utilis\u00e9s dans le paquetage \u00e9taient \u00e0 l'origine inclus dans le paquetage, \u00e0 partir de la version 0.8.0, ils ont \u00e9t\u00e9 publi\u00e9s dans un paquetage R s\u00e9par\u00e9 sigmo\u00efde, dans le but de permettre une utilisation plus g\u00e9n\u00e9rale.", "token2charspan": [[0, 3], [4, 13], [14, 23], [24, 26], [27, 30], [31, 38], [39, 47], [48, 52], [53, 55], [56, 65], [66, 73], [74, 75], [76, 78], [78, 85], [86, 92], [93, 97], [98, 100], [101, 110], [110, 111], [112, 113], [114, 120], [121, 123], [124, 126], [127, 134], [135, 140], [140, 141], [142, 145], [146, 149], [150, 153], [154, 161], [162, 166], [167, 169], [170, 179], [180, 181], [182, 188], [189, 197], [197, 198], [199, 203], [204, 206], [207, 210], [211, 213], [214, 223], [224, 227], [228, 239], [240, 244], [245, 253], [253, 254]]}
{"doc_key": "ai-train-71", "ner": [[1, 1, "programlang"], [8, 12, "organisation"], [14, 14, "organisation"], [22, 22, "location"], [24, 24, "location"], [27, 28, "researcher"], [30, 31, "researcher"], [33, 34, "researcher"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "relations": [[1, 1, 27, 28, "artifact", "", true, false], [1, 1, 30, 31, "artifact", "", true, false], [1, 1, 33, 34, "artifact", "", true, false], [14, 14, 8, 12, "named", "", false, false], [14, 14, 22, 22, "physical", "", false, false], [22, 22, 24, 24, "physical", "", false, false], [27, 28, 8, 12, "role", "", false, false], [30, 31, 8, 12, "role", "", false, false], [33, 34, 8, 12, "role", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "sentence": ["Le", "logo", "a", "\u00e9t\u00e9", "cr\u00e9\u00e9", "en", "1967", "chez", "Bolt", ",", "Beranek", "et", "Newman", "(", "BBN", ")", ",", "un", "cabinet", "de", "recherche", "de", "Cambridge", ",", "Massachusetts", ",", "par", "Wally", "Feurzeig", ",", "Cynthia", "Solomon", "et", "Seymour", "Papert", "."], "sentence-detokenized": "Le logo a \u00e9t\u00e9 cr\u00e9\u00e9 en 1967 chez Bolt, Beranek et Newman (BBN), un cabinet de recherche de Cambridge, Massachusetts, par Wally Feurzeig, Cynthia Solomon et Seymour Papert.", "token2charspan": [[0, 2], [3, 7], [8, 9], [10, 13], [14, 18], [19, 21], [22, 26], [27, 31], [32, 36], [36, 37], [38, 45], [46, 48], [49, 55], [56, 57], [57, 60], [60, 61], [61, 62], [63, 65], [66, 73], [74, 76], [77, 86], [87, 89], [90, 99], [99, 100], [101, 114], [114, 115], [116, 119], [120, 125], [126, 134], [134, 135], [136, 143], [144, 151], [152, 154], [155, 162], [163, 169], [169, 170]]}
{"doc_key": "ai-train-72", "ner": [[0, 1, "misc"], [12, 14, "field"], [25, 26, "field"], [30, 32, "algorithm"], [35, 36, "algorithm"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[0, 1, 12, 14, "part-of", "", false, false], [0, 1, 25, 26, "compare", "", false, false], [30, 32, 25, 26, "part-of", "", false, false], [35, 36, 25, 26, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["La", "neuro\u00e9volution", "est", "couramment", "utilis\u00e9e", "dans", "le", "cadre", "du", "paradigme", "de", "l'", "apprentissage", "par", "renforcement", ",", "et", "elle", "peut", "\u00eatre", "compar\u00e9e", "aux", "techniques", "conventionnelles", "d'", "apprentissage", "profond", "qui", "utilisent", "la", "descente", "de", "gradient", "sur", "un", "r\u00e9seau", "neuronal", "\u00e0", "topologie", "fixe", "."], "sentence-detokenized": "La neuro\u00e9volution est couramment utilis\u00e9e dans le cadre du paradigme de l'apprentissage par renforcement, et elle peut \u00eatre compar\u00e9e aux techniques conventionnelles d'apprentissage profond qui utilisent la descente de gradient sur un r\u00e9seau neuronal \u00e0 topologie fixe.", "token2charspan": [[0, 2], [3, 17], [18, 21], [22, 32], [33, 41], [42, 46], [47, 49], [50, 55], [56, 58], [59, 68], [69, 71], [72, 74], [74, 87], [88, 91], [92, 104], [104, 105], [106, 108], [109, 113], [114, 118], [119, 123], [124, 132], [133, 136], [137, 147], [148, 164], [165, 167], [167, 180], [181, 188], [189, 192], [193, 202], [203, 205], [206, 214], [215, 217], [218, 226], [227, 230], [231, 233], [234, 240], [241, 249], [250, 251], [252, 261], [262, 266], [266, 267]]}
{"doc_key": "ai-train-73", "ner": [[4, 5, "algorithm"], [57, 59, "metrics"], [61, 61, "metrics"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[61, 61, 57, 59, "named", "", false, false]], "relations_mapping_to_source": [0], "sentence": ["Si", "nous", "utilisons", "les", "moindres", "carr\u00e9s", "pour", "ajuster", "une", "fonction", "sous", "la", "forme", "d'", "un", "hyperplan", "\u0177", "=", "a", "+", "\u03b2", "supT", "/", "sup", "x", "aux", "donn\u00e9es", "(", "x", "sub", "i", "/", "sub", ",", "y", "sub", "i", "/", "sub", ")", "sub", "1", "\u2264", "i", "\u2264n", "/", "sub", ",", "nous", "pourrions", "alors", "\u00e9valuer", "l'", "ajustement", "en", "utilisant", "l'", "erreur", "quadratique", "moyenne", "(", "EQM", ")", "."], "sentence-detokenized": "Si nous utilisons les moindres carr\u00e9s pour ajuster une fonction sous la forme d'un hyperplan \u0177 = a + \u03b2 supT / sup x aux donn\u00e9es (x sub i / sub, y sub i / sub) sub 1 \u2264 i \u2264n / sub, nous pourrions alors \u00e9valuer l'ajustement en utilisant l'erreur quadratique moyenne (EQM).", "token2charspan": [[0, 2], [3, 7], [8, 17], [18, 21], [22, 30], [31, 37], [38, 42], [43, 50], [51, 54], [55, 63], [64, 68], [69, 71], [72, 77], [78, 80], [80, 82], [83, 92], [93, 94], [95, 96], [97, 98], [99, 100], [101, 102], [103, 107], [108, 109], [110, 113], [114, 115], [116, 119], [120, 127], [128, 129], [129, 130], [131, 134], [135, 136], [137, 138], [139, 142], [142, 143], [144, 145], [146, 149], [150, 151], [152, 153], [154, 157], [157, 158], [159, 162], [163, 164], [165, 166], [167, 168], [169, 171], [172, 173], [174, 177], [177, 178], [179, 183], [184, 193], [194, 199], [200, 207], [208, 210], [210, 220], [221, 223], [224, 233], [234, 236], [236, 242], [243, 254], [255, 262], [263, 264], [264, 267], [267, 268], [268, 269]]}
{"doc_key": "ai-train-74", "ner": [[7, 7, "country"], [9, 10, "country"], [13, 13, "country"], [16, 16, "country"], [19, 19, "country"], [22, 22, "country"], [25, 25, "country"], [28, 28, "country"], [31, 31, "country"], [34, 34, "country"], [37, 37, "country"], [40, 40, "country"], [43, 43, "country"], [46, 46, "country"], [49, 49, "country"], [52, 54, "country"], [57, 57, "country"], [60, 60, "country"], [63, 63, "country"], [66, 66, "country"], [69, 69, "country"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20], "relations": [], "relations_mapping_to_source": [], "sentence": ["L'", "entreprise", "dispose", "de", "sites", "internationaux", "en", "Australie", ",", "au", "Br\u00e9sil", ",", "au", "Canada", ",", "en", "Chine", ",", "en", "Allemagne", ",", "en", "Inde", ",", "en", "Italie", ",", "au", "Japon", ",", "en", "Cor\u00e9e", ",", "en", "Lituanie", ",", "en", "Pologne", ",", "en", "Malaisie", ",", "aux", "Philippines", ",", "en", "Russie", ",", "\u00e0", "Singapour", ",", "en", "Afrique", "du", "Sud", ",", "en", "Espagne", ",", "\u00e0", "Ta\u00efwan", ",", "en", "Tha\u00eflande", ",", "en", "Turquie", "et", "au", "Royaume-Uni", "."], "sentence-detokenized": "L'entreprise dispose de sites internationaux en Australie, au Br\u00e9sil, au Canada, en Chine, en Allemagne, en Inde, en Italie, au Japon, en Cor\u00e9e, en Lituanie, en Pologne, en Malaisie, aux Philippines, en Russie, \u00e0 Singapour, en Afrique du Sud, en Espagne, \u00e0 Ta\u00efwan, en Tha\u00eflande, en Turquie et au Royaume-Uni.", "token2charspan": [[0, 2], [2, 12], [13, 20], [21, 23], [24, 29], [30, 44], [45, 47], [48, 57], [57, 58], [59, 61], [62, 68], [68, 69], [70, 72], [73, 79], [79, 80], [81, 83], [84, 89], [89, 90], [91, 93], [94, 103], [103, 104], [105, 107], [108, 112], [112, 113], [114, 116], [117, 123], [123, 124], [125, 127], [128, 133], [133, 134], [135, 137], [138, 143], [143, 144], [145, 147], [148, 156], [156, 157], [158, 160], [161, 168], [168, 169], [170, 172], [173, 181], [181, 182], [183, 186], [187, 198], [198, 199], [200, 202], [203, 209], [209, 210], [211, 212], [213, 222], [222, 223], [224, 226], [227, 234], [235, 237], [238, 241], [241, 242], [243, 245], [246, 253], [253, 254], [255, 256], [257, 263], [263, 264], [265, 267], [268, 277], [277, 278], [279, 281], [282, 289], [290, 292], [293, 295], [296, 307], [307, 308]]}
{"doc_key": "ai-train-75", "ner": [[3, 3, "misc"], [6, 9, "field"], [15, 15, "organisation"], [19, 26, "university"], [32, 34, "organisation"], [38, 43, "university"], [50, 51, "university"], [55, 56, "university"], [60, 62, "university"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "relations": [[3, 3, 6, 9, "topic", "", false, false], [3, 3, 15, 15, "origin", "", false, false], [3, 3, 19, 26, "origin", "", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["Titulaire", "d'", "un", "D.Sc", ".", "en", "g\u00e9nie", "\u00e9lectrique", "et", "informatique", "(", "2000", ")", "de", "l'", "Inria", "et", "de", "l'", "Universit\u00e9", "de", "Nice", "Sophia", "Antipolis", ",", "il", "a", "occup\u00e9", "des", "postes", "permanents", "chez", "Siemens", "Corporate", "Technology", ",", "\u00e0", "l'", "\u00c9cole", "des", "ponts", "ParisTech", "ainsi", "que", "des", "postes", "de", "visiteur", "\u00e0", "l'", "Universit\u00e9", "Rutgers", ",", "\u00e0", "l'", "Universit\u00e9", "Yale", "et", "\u00e0", "l'", "Universit\u00e9", "de", "Houston", "."], "sentence-detokenized": "Titulaire d'un D.Sc. en g\u00e9nie \u00e9lectrique et informatique (2000) de l'Inria et de l'Universit\u00e9 de Nice Sophia Antipolis, il a occup\u00e9 des postes permanents chez Siemens Corporate Technology, \u00e0 l'\u00c9cole des ponts ParisTech ainsi que des postes de visiteur \u00e0 l'Universit\u00e9 Rutgers, \u00e0 l'Universit\u00e9 Yale et \u00e0 l'Universit\u00e9 de Houston.", "token2charspan": [[0, 9], [10, 12], [12, 14], [15, 19], [19, 20], [21, 23], [24, 29], [30, 40], [41, 43], [44, 56], [57, 58], [58, 62], [62, 63], [64, 66], [67, 69], [69, 74], [75, 77], [78, 80], [81, 83], [83, 93], [94, 96], [97, 101], [102, 108], [109, 118], [118, 119], [120, 122], [123, 124], [125, 131], [132, 135], [136, 142], [143, 153], [154, 158], [159, 166], [167, 176], [177, 187], [187, 188], [189, 190], [191, 193], [193, 198], [199, 202], [203, 208], [209, 218], [219, 224], [225, 228], [229, 232], [233, 239], [240, 242], [243, 251], [252, 253], [254, 256], [256, 266], [267, 274], [274, 275], [276, 277], [278, 280], [280, 290], [291, 295], [296, 298], [299, 300], [301, 303], [303, 313], [314, 316], [317, 324], [324, 325]]}
{"doc_key": "ai-train-76", "ner": [[11, 12, "researcher"], [14, 14, "researcher"], [19, 20, "product"], [22, 22, "country"], [25, 25, "product"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[14, 14, 11, 12, "role", "licensing_patent_to", false, false], [14, 14, 22, 22, "physical", "", false, false], [25, 25, 14, 14, "artifact", "", false, false], [25, 25, 19, 20, "type-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["D\u00e9tenteur", "d'", "une", "licence", "du", "brevet", "original", "accord\u00e9", "\u00e0", "l'", "inventeur", "George", "Devol", ",", "Engelberger", "a", "d\u00e9velopp\u00e9", "le", "premier", "robot", "industriel", "aux", "\u00c9tats-Unis", ",", "l'", "Unimate", ",", "dans", "les", "ann\u00e9es", "1950", "."], "sentence-detokenized": "D\u00e9tenteur d'une licence du brevet original accord\u00e9 \u00e0 l'inventeur George Devol, Engelberger a d\u00e9velopp\u00e9 le premier robot industriel aux \u00c9tats-Unis, l'Unimate, dans les ann\u00e9es 1950.", "token2charspan": [[0, 9], [10, 12], [12, 15], [16, 23], [24, 26], [27, 33], [34, 42], [43, 50], [51, 52], [53, 55], [55, 64], [65, 71], [72, 77], [77, 78], [79, 90], [91, 92], [93, 102], [103, 105], [106, 113], [114, 119], [120, 130], [131, 134], [135, 145], [145, 146], [147, 149], [149, 156], [156, 157], [158, 162], [163, 166], [167, 173], [174, 178], [178, 179]]}
{"doc_key": "ai-train-77", "ner": [[4, 5, "task"], [11, 12, "task"]], "ner_mapping_to_source": [0, 1], "relations": [], "relations_mapping_to_source": [], "sentence": ["L'", "entr\u00e9e", "est", "appel\u00e9e", "reconnaissance", "vocale", "et", "la", "sortie", "est", "appel\u00e9e", "synth\u00e8se", "vocale", "."], "sentence-detokenized": "L'entr\u00e9e est appel\u00e9e reconnaissance vocale et la sortie est appel\u00e9e synth\u00e8se vocale.", "token2charspan": [[0, 2], [2, 8], [9, 12], [13, 20], [21, 35], [36, 42], [43, 45], [46, 48], [49, 55], [56, 59], [60, 67], [68, 76], [77, 83], [83, 84]]}
{"doc_key": "ai-train-78", "ner": [[5, 5, "programlang"], [8, 8, "programlang"], [12, 12, "programlang"], [19, 19, "programlang"], [31, 31, "programlang"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[5, 5, 12, 12, "named", "", false, false], [8, 8, 5, 5, "origin", "descendant_of", false, false], [8, 8, 19, 19, "general-affiliation", "", false, false], [8, 8, 31, 31, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Parmi", "les", "descendants", "du", "langage", "CLIPS", ",", "citons", "Jess", "(", "partie", "de", "CLIPS", "bas\u00e9e", "sur", "les", "r\u00e8gles", "r\u00e9\u00e9crite", "en", "Java", ",", "elle", "a", "ensuite", "\u00e9volu\u00e9", "dans", "une", "autre", "direction", ")", ",", "JESS", "a", "\u00e9t\u00e9", "inspir\u00e9", "\u00e0", "l'", "origine", "de"], "sentence-detokenized": "Parmi les descendants du langage CLIPS, citons Jess (partie de CLIPS bas\u00e9e sur les r\u00e8gles r\u00e9\u00e9crite en Java, elle a ensuite \u00e9volu\u00e9 dans une autre direction), JESS a \u00e9t\u00e9 inspir\u00e9 \u00e0 l'origine de", "token2charspan": [[0, 5], [6, 9], [10, 21], [22, 24], [25, 32], [33, 38], [38, 39], [40, 46], [47, 51], [52, 53], [53, 59], [60, 62], [63, 68], [69, 74], [75, 78], [79, 82], [83, 89], [90, 98], [99, 101], [102, 106], [106, 107], [108, 112], [113, 114], [115, 122], [123, 129], [130, 134], [135, 138], [139, 144], [145, 154], [154, 155], [155, 156], [157, 161], [162, 163], [164, 167], [168, 175], [176, 177], [178, 180], [180, 187], [188, 190]]}
{"doc_key": "ai-train-79", "ner": [[6, 6, "product"], [13, 16, "product"], [19, 20, "organisation"], [24, 25, "product"], [47, 49, "product"], [52, 55, "product"], [79, 81, "misc"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "relations": [[13, 16, 6, 6, "type-of", "", false, false], [19, 20, 13, 16, "usage", "", false, false], [24, 25, 19, 20, "artifact", "", false, false], [47, 49, 19, 20, "origin", "", true, false], [47, 49, 79, 81, "related-to", "", true, false], [52, 55, 19, 20, "origin", "", true, false], [52, 55, 79, 81, "related-to", "", true, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "sentence": ["Elle", "a", "\u00e9galement", "cr\u00e9\u00e9", "des", "applications", "AGV", "intelligentes", "flexibles", ",", "en", "concevant", "le", "syst\u00e8me", "de", "commande", "Motivity", "utilis\u00e9", "par", "RMT", "Robotics", "pour", "d\u00e9velopper", "son", "ADAM", "iAGV", "(", "Self", "-Guided", "Vehicle", ")", ",", "utilis\u00e9", "pour", "des", "op\u00e9rations", "complexes", "de", "pr\u00e9l\u00e8vement", "et", "de", "placement", ",", "en", "conjonction", "avec", "des", "syst\u00e8mes", "de", "portique", "et", "des", "bras", "de", "robot", "industriel", ",", "utilis\u00e9s", "dans", "des", "usines", "de", "fournitures", "automobiles", "de", "premier", "rang", "pour", "d\u00e9placer", "des", "produits", "d'", "un", "processus", "\u00e0", "l'", "autre", "dans", "des", "agencements", "non", "lin\u00e9aires", "."], "sentence-detokenized": "Elle a \u00e9galement cr\u00e9\u00e9 des applications AGV intelligentes flexibles, en concevant le syst\u00e8me de commande Motivity utilis\u00e9 par RMT Robotics pour d\u00e9velopper son ADAM iAGV (Self-Guided Vehicle), utilis\u00e9 pour des op\u00e9rations complexes de pr\u00e9l\u00e8vement et de placement, en conjonction avec des syst\u00e8mes de portique et des bras de robot industriel, utilis\u00e9s dans des usines de fournitures automobiles de premier rang pour d\u00e9placer des produits d'un processus \u00e0 l'autre dans des agencements non lin\u00e9aires.", "token2charspan": [[0, 4], [5, 6], [7, 16], [17, 21], [22, 25], [26, 38], [39, 42], [43, 56], [57, 66], [66, 67], [68, 70], [71, 80], [81, 83], [84, 91], [92, 94], [95, 103], [104, 112], [113, 120], [121, 124], [125, 128], [129, 137], [138, 142], [143, 153], [154, 157], [158, 162], [163, 167], [168, 169], [169, 173], [173, 180], [181, 188], [188, 189], [189, 190], [191, 198], [199, 203], [204, 207], [208, 218], [219, 228], [229, 231], [232, 243], [244, 246], [247, 249], [250, 259], [259, 260], [261, 263], [264, 275], [276, 280], [281, 284], [285, 293], [294, 296], [297, 305], [306, 308], [309, 312], [313, 317], [318, 320], [321, 326], [327, 337], [337, 338], [339, 347], [348, 352], [353, 356], [357, 363], [364, 366], [367, 378], [379, 390], [391, 393], [394, 401], [402, 406], [407, 411], [412, 420], [421, 424], [425, 433], [434, 436], [436, 438], [439, 448], [449, 450], [451, 453], [453, 458], [459, 463], [464, 467], [468, 479], [480, 483], [484, 493], [493, 494]]}
{"doc_key": "ai-train-80", "ner": [[7, 9, "metrics"]], "ner_mapping_to_source": [0], "relations": [], "relations_mapping_to_source": [], "sentence": ["Les", "param\u00e8tres", "\u03b2", "sont", "g\u00e9n\u00e9ralement", "estim\u00e9s", "par", "maximum", "de", "vraisemblance", "."], "sentence-detokenized": "Les param\u00e8tres \u03b2 sont g\u00e9n\u00e9ralement estim\u00e9s par maximum de vraisemblance.", "token2charspan": [[0, 3], [4, 14], [15, 16], [17, 21], [22, 34], [35, 42], [43, 46], [47, 54], [55, 57], [58, 71], [71, 72]]}
{"doc_key": "ai-train-81", "ner": [[3, 5, "task"], [9, 9, "metrics"], [12, 12, "metrics"], [15, 15, "metrics"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[9, 9, 3, 5, "part-of", "", false, false], [12, 12, 3, 5, "part-of", "", false, false], [15, 15, 3, 5, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["Les", "m\u00e9triques", "de", "recherche", "d'", "information", "telles", "que", "la", "pr\u00e9cision", "et", "le", "rappel", "ou", "le", "DCG", "sont", "utiles", "pour", "\u00e9valuer", "la", "qualit\u00e9", "d'", "une", "m\u00e9thode", "de", "recommandation", "."], "sentence-detokenized": "Les m\u00e9triques de recherche d'information telles que la pr\u00e9cision et le rappel ou le DCG sont utiles pour \u00e9valuer la qualit\u00e9 d'une m\u00e9thode de recommandation.", "token2charspan": [[0, 3], [4, 13], [14, 16], [17, 26], [27, 29], [29, 40], [41, 47], [48, 51], [52, 54], [55, 64], [65, 67], [68, 70], [71, 77], [78, 80], [81, 83], [84, 87], [88, 92], [93, 99], [100, 104], [105, 112], [113, 115], [116, 123], [124, 126], [126, 129], [130, 137], [138, 140], [141, 155], [155, 156]]}
{"doc_key": "ai-train-82", "ner": [[7, 8, "product"]], "ner_mapping_to_source": [0], "relations": [], "relations_mapping_to_source": [], "sentence": ["Une", "usine", "typique", "contient", "des", "centaines", "de", "robots", "industriels", "travaillant", "sur", "des", "lignes", "de", "production", "enti\u00e8rement", "automatis\u00e9es", ",", "avec", "un", "robot", "pour", "dix", "travailleurs", "humains", "."], "sentence-detokenized": "Une usine typique contient des centaines de robots industriels travaillant sur des lignes de production enti\u00e8rement automatis\u00e9es, avec un robot pour dix travailleurs humains.", "token2charspan": [[0, 3], [4, 9], [10, 17], [18, 26], [27, 30], [31, 40], [41, 43], [44, 50], [51, 62], [63, 74], [75, 78], [79, 82], [83, 89], [90, 92], [93, 103], [104, 115], [116, 128], [128, 129], [130, 134], [135, 137], [138, 143], [144, 148], [149, 152], [153, 165], [166, 173], [173, 174]]}
{"doc_key": "ai-train-83", "ner": [[8, 10, "product"], [16, 18, "field"], [22, 24, "task"], [27, 29, "task"], [32, 34, "task"], [37, 39, "task"], [42, 44, "task"], [47, 49, "task"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "relations": [[16, 18, 8, 10, "usage", "", false, true], [22, 24, 16, 18, "part-of", "", false, false], [27, 29, 16, 18, "part-of", "", false, false], [32, 34, 16, 18, "part-of", "", false, false], [37, 39, 16, 18, "part-of", "", false, false], [42, 44, 16, 18, "part-of", "", false, false], [47, 49, 16, 18, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "sentence": ["Au", "cours", "de", "la", "derni\u00e8re", "d\u00e9cennie", ",", "les", "PCNN", "ont", "\u00e9t\u00e9", "utilis\u00e9s", "dans", "diverses", "applications", "de", "traitement", "d'", "images", ",", "notamment", "la", "segmentation", "d'", "images", ",", "la", "g\u00e9n\u00e9ration", "de", "caract\u00e9ristiques", ",", "l'", "extraction", "de", "visages", ",", "la", "d\u00e9tection", "de", "mouvements", ",", "la", "croissance", "de", "r\u00e9gions", "et", "la", "r\u00e9duction", "du", "bruit", "."], "sentence-detokenized": "Au cours de la derni\u00e8re d\u00e9cennie, les PCNN ont \u00e9t\u00e9 utilis\u00e9s dans diverses applications de traitement d'images, notamment la segmentation d'images, la g\u00e9n\u00e9ration de caract\u00e9ristiques, l'extraction de visages, la d\u00e9tection de mouvements, la croissance de r\u00e9gions et la r\u00e9duction du bruit.", "token2charspan": [[0, 2], [3, 8], [9, 11], [12, 14], [15, 23], [24, 32], [32, 33], [34, 37], [38, 42], [43, 46], [47, 50], [51, 59], [60, 64], [65, 73], [74, 86], [87, 89], [90, 100], [101, 103], [103, 109], [109, 110], [111, 120], [121, 123], [124, 136], [137, 139], [139, 145], [145, 146], [147, 149], [150, 160], [161, 163], [164, 180], [180, 181], [182, 184], [184, 194], [195, 197], [198, 205], [205, 206], [207, 209], [210, 219], [220, 222], [223, 233], [233, 234], [235, 237], [238, 248], [249, 251], [252, 259], [260, 262], [263, 265], [266, 275], [276, 278], [279, 284], [284, 285]]}
{"doc_key": "ai-train-84", "ner": [[0, 0, "researcher"], [19, 21, "field"], [26, 30, "misc"], [32, 41, "conference"], [43, 43, "conference"], [48, 51, "misc"], [53, 59, "conference"], [60, 61, "conference"], [63, 67, "conference"], [69, 69, "conference"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "relations": [[0, 0, 19, 21, "related-to", "contributes_to", false, false], [0, 0, 26, 30, "win-defeat", "", false, false], [0, 0, 48, 51, "win-defeat", "", false, false], [26, 30, 32, 41, "temporal", "", false, false], [43, 43, 32, 41, "named", "", false, false], [48, 51, 53, 59, "temporal", "", false, false], [48, 51, 63, 67, "temporal", "", false, false], [60, 61, 53, 59, "named", "", false, false], [69, 69, 63, 67, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "sentence": ["Xu", "a", "publi\u00e9", "plus", "de", "50", "articles", "dans", "des", "conf\u00e9rences", "internationales", "et", "des", "revues", "dans", "le", "domaine", "de", "la", "vision", "par", "ordinateur", "et", "a", "remport\u00e9", "le", "prix", "du", "meilleur", "article", "\u00e0", "la", "conf\u00e9rence", "internationale", "sur", "le", "rendu", "et", "l'", "animation", "non", "photor\u00e9alistes", "(", "NPAR", ")", "2012", "et", "le", "prix", "du", "meilleur", "r\u00e9viseur", "aux", "conf\u00e9rences", "internationales", "Asian", "Conference", "on", "Computer", "Vision", "ACCV", "2012", "et", "International", "Conference", "on", "Computer", "Vision", "(", "ICCV", ")", "2015", "."], "sentence-detokenized": "Xu a publi\u00e9 plus de 50 articles dans des conf\u00e9rences internationales et des revues dans le domaine de la vision par ordinateur et a remport\u00e9 le prix du meilleur article \u00e0 la conf\u00e9rence internationale sur le rendu et l'animation non photor\u00e9alistes (NPAR) 2012 et le prix du meilleur r\u00e9viseur aux conf\u00e9rences internationales Asian Conference on Computer Vision ACCV 2012 et International Conference on Computer Vision (ICCV) 2015.", "token2charspan": [[0, 2], [3, 4], [5, 11], [12, 16], [17, 19], [20, 22], [23, 31], [32, 36], [37, 40], [41, 52], [53, 68], [69, 71], [72, 75], [76, 82], [83, 87], [88, 90], [91, 98], [99, 101], [102, 104], [105, 111], [112, 115], [116, 126], [127, 129], [130, 131], [132, 140], [141, 143], [144, 148], [149, 151], [152, 160], [161, 168], [169, 170], [171, 173], [174, 184], [185, 199], [200, 203], [204, 206], [207, 212], [213, 215], [216, 218], [218, 227], [228, 231], [232, 246], [247, 248], [248, 252], [252, 253], [254, 258], [259, 261], [262, 264], [265, 269], [270, 272], [273, 281], [282, 290], [291, 294], [295, 306], [307, 322], [323, 328], [329, 339], [340, 342], [343, 351], [352, 358], [359, 363], [364, 368], [369, 371], [372, 385], [386, 396], [397, 399], [400, 408], [409, 415], [416, 417], [417, 421], [421, 422], [423, 427], [427, 428]]}
{"doc_key": "ai-train-85", "ner": [[0, 0, "programlang"], [2, 2, "field"], [5, 6, "field"], [9, 10, "misc"], [18, 19, "researcher"], [14, 16, "misc"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5], "relations": [[0, 0, 2, 2, "part-of", "", false, false], [0, 0, 5, 6, "part-of", "", false, false], [0, 0, 9, 10, "type-of", "", false, false], [14, 16, 0, 0, "usage", "", false, false], [14, 16, 18, 19, "origin", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4], "sentence": ["CycL", "en", "informatique", "et", "en", "intelligence", "artificielle", "est", "un", "langage", "ontologique", "utilis\u00e9", "par", "le", "projet", "Cyc", "artificial", "de", "Doug", "Lenat", "."], "sentence-detokenized": "CycL en informatique et en intelligence artificielle est un langage ontologique utilis\u00e9 par le projet Cyc artificial de Doug Lenat.", "token2charspan": [[0, 4], [5, 7], [8, 20], [21, 23], [24, 26], [27, 39], [40, 52], [53, 56], [57, 59], [60, 67], [68, 79], [80, 87], [88, 91], [92, 94], [95, 101], [102, 105], [106, 116], [117, 119], [120, 124], [125, 130], [130, 131]]}
{"doc_key": "ai-train-86", "ner": [[3, 5, "task"], [8, 10, "metrics"], [14, 18, "metrics"], [20, 24, "metrics"], [32, 33, "misc"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[8, 10, 3, 5, "part-of", "", false, false], [14, 18, 8, 10, "named", "", false, false], [20, 24, 8, 10, "named", "", false, false], [32, 33, 8, 10, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Toujours", "dans", "l'", "analyse", "de", "r\u00e9gression", ",", "l'", "erreur", "quadratique", "moyenne", ",", "souvent", "appel\u00e9e", "erreur", "quadratique", "moyenne", "de", "pr\u00e9diction", "ou", "erreur", "quadratique", "moyenne", "hors", "\u00e9chantillon", ",", "peut", "d\u00e9signer", "la", "valeur", "moyenne", "des", "\u00e9carts", "quadratiques", "des", "pr\u00e9dictions", "par", "rapport", "aux", "valeurs", "VRAIES", ",", "sur", "un", "espace", "de", "test", "hors", "\u00e9chantillon", ",", "g\u00e9n\u00e9r\u00e9", "par", "un", "mod\u00e8le", "estim\u00e9", "sur", "un", "espace", "d'", "\u00e9chantillon", "particulier", "."], "sentence-detokenized": "Toujours dans l'analyse de r\u00e9gression, l'erreur quadratique moyenne, souvent appel\u00e9e erreur quadratique moyenne de pr\u00e9diction ou erreur quadratique moyenne hors \u00e9chantillon, peut d\u00e9signer la valeur moyenne des \u00e9carts quadratiques des pr\u00e9dictions par rapport aux valeurs VRAIES, sur un espace de test hors \u00e9chantillon, g\u00e9n\u00e9r\u00e9 par un mod\u00e8le estim\u00e9 sur un espace d'\u00e9chantillon particulier.", "token2charspan": [[0, 8], [9, 13], [14, 16], [16, 23], [24, 26], [27, 37], [37, 38], [39, 41], [41, 47], [48, 59], [60, 67], [67, 68], [69, 76], [77, 84], [85, 91], [92, 103], [104, 111], [112, 114], [115, 125], [126, 128], [129, 135], [136, 147], [148, 155], [156, 160], [161, 172], [172, 173], [174, 178], [179, 187], [188, 190], [191, 197], [198, 205], [206, 209], [210, 216], [217, 229], [230, 233], [234, 245], [246, 249], [250, 257], [258, 261], [262, 269], [270, 276], [276, 277], [278, 281], [282, 284], [285, 291], [292, 294], [295, 299], [300, 304], [305, 316], [316, 317], [318, 324], [325, 328], [329, 331], [332, 338], [339, 345], [346, 349], [350, 352], [353, 359], [360, 362], [362, 373], [374, 385], [385, 386]]}
{"doc_key": "ai-train-87", "ner": [[11, 11, "algorithm"], [13, 13, "algorithm"], [20, 21, "algorithm"], [37, 39, "metrics"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[11, 11, 13, 13, "compare", "", false, false], [11, 11, 20, 21, "named", "", false, false]], "relations_mapping_to_source": [0, 1], "sentence": ["En", "ce", "qui", "concerne", "les", "r\u00e9sultats", ",", "les", "descripteurs", "de", "blocs", "C-HOG", "et", "R-HOG", "ont", "des", "performances", "comparables", ",", "les", "descripteurs", "C-HOG", "conservant", "un", "l\u00e9ger", "avantage", "dans", "le", "taux", "d'", "\u00e9chec", "de", "la", "d\u00e9tection", "pour", "des", "taux", "de", "faux", "positifs", "fixes", "sur", "les", "deux", "ensembles", "de", "donn\u00e9es", "."], "sentence-detokenized": "En ce qui concerne les r\u00e9sultats, les descripteurs de blocs C-HOG et R-HOG ont des performances comparables, les descripteurs C-HOG conservant un l\u00e9ger avantage dans le taux d'\u00e9chec de la d\u00e9tection pour des taux de faux positifs fixes sur les deux ensembles de donn\u00e9es.", "token2charspan": [[0, 2], [3, 5], [6, 9], [10, 18], [19, 22], [23, 32], [32, 33], [34, 37], [38, 50], [51, 53], [54, 59], [60, 65], [66, 68], [69, 74], [75, 78], [79, 82], [83, 95], [96, 107], [107, 108], [109, 112], [113, 125], [126, 131], [132, 142], [143, 145], [146, 151], [152, 160], [161, 165], [166, 168], [169, 173], [174, 176], [176, 181], [182, 184], [185, 187], [188, 197], [198, 202], [203, 206], [207, 211], [212, 214], [215, 219], [220, 228], [229, 234], [235, 238], [239, 242], [243, 247], [248, 257], [258, 260], [261, 268], [268, 269]]}
{"doc_key": "ai-train-88", "ner": [[9, 12, "algorithm"], [15, 16, "misc"], [19, 21, "algorithm"], [24, 25, "algorithm"], [28, 30, "algorithm"], [33, 36, "algorithm"], [39, 42, "algorithm"], [45, 46, "misc"], [50, 53, "algorithm"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8], "relations": [[9, 12, 15, 16, "usage", "", false, false], [19, 21, 45, 46, "usage", "", false, false], [24, 25, 45, 46, "usage", "", false, false], [28, 30, 45, 46, "usage", "", false, false], [33, 36, 45, 46, "usage", "", false, false], [39, 42, 45, 46, "usage", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5], "sentence": ["Les", "algorithmes", "de", "reconnaissance", "les", "plus", "r\u00e9pandus", "sont", "l'", "analyse", "en", "composantes", "principales", "utilisant", "les", "faces", "propres", ",", "l'", "analyse", "discriminante", "lin\u00e9aire", ",", "l'", "appariement", "\u00e9lastique", "utilisant", "l'", "algorithme", "de", "Fisherface", ",", "le", "mod\u00e8le", "de", "Markov", "cach\u00e9", ",", "l'", "apprentissage", "du", "sous-espace", "multilin\u00e9aire", "utilisant", "la", "repr\u00e9sentation", "tensorielle", ",", "et", "l'", "appariement", "dynamique", "des", "liens", "motiv\u00e9", "par", "les", "neurones", "."], "sentence-detokenized": "Les algorithmes de reconnaissance les plus r\u00e9pandus sont l'analyse en composantes principales utilisant les faces propres, l'analyse discriminante lin\u00e9aire, l'appariement \u00e9lastique utilisant l'algorithme de Fisherface, le mod\u00e8le de Markov cach\u00e9, l'apprentissage du sous-espace multilin\u00e9aire utilisant la repr\u00e9sentation tensorielle, et l'appariement dynamique des liens motiv\u00e9 par les neurones.", "token2charspan": [[0, 3], [4, 15], [16, 18], [19, 33], [34, 37], [38, 42], [43, 51], [52, 56], [57, 59], [59, 66], [67, 69], [70, 81], [82, 93], [94, 103], [104, 107], [108, 113], [114, 121], [121, 122], [123, 125], [125, 132], [133, 146], [147, 155], [155, 156], [157, 159], [159, 170], [171, 180], [181, 190], [191, 193], [193, 203], [204, 206], [207, 217], [217, 218], [219, 221], [222, 228], [229, 231], [232, 238], [239, 244], [244, 245], [246, 248], [248, 261], [262, 264], [265, 276], [277, 290], [291, 300], [301, 303], [304, 318], [319, 330], [330, 331], [332, 334], [335, 337], [337, 348], [349, 358], [359, 362], [363, 368], [369, 375], [376, 379], [380, 383], [384, 392], [392, 393]]}
{"doc_key": "ai-train-89", "ner": [[4, 12, "misc"], [27, 29, "location"], [48, 50, "location"], [67, 67, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[27, 29, 4, 12, "temporal", "", false, false], [48, 50, 4, 12, "temporal", "", false, false]], "relations_mapping_to_source": [0, 1], "sentence": ["\u00c0", "partir", "de", "l'", "\u00e9dition", "2019", "du", "Festival", "international", "du", "film", "de", "Toronto", ",", "il", "se", "peut", "d\u00e9sormais", "que", "les", "films", "ne", "puissent", "plus", "\u00eatre", "projet\u00e9s", "au", "Scotiabank", "Theatre", "Toronto", "-", "l'", "un", "des", "principaux", "sites", "du", "festival", "-", "et", "qu'", "ils", "soient", "projet\u00e9s", "ailleurs", "(", "notamment", "au", "TIFF", "Bell", "Lightbox", "et", "dans", "d'", "autres", "cin\u00e9mas", "locaux", ")", "s'", "ils", "sont", "distribu\u00e9s", "par", "un", "service", "tel", "que", "Netflix", "."], "sentence-detokenized": "\u00c0 partir de l'\u00e9dition 2019 du Festival international du film de Toronto, il se peut d\u00e9sormais que les films ne puissent plus \u00eatre projet\u00e9s au Scotiabank Theatre Toronto - l'un des principaux sites du festival - et qu'ils soient projet\u00e9s ailleurs (notamment au TIFF Bell Lightbox et dans d'autres cin\u00e9mas locaux) s'ils sont distribu\u00e9s par un service tel que Netflix.", "token2charspan": [[0, 1], [2, 8], [9, 11], [12, 14], [14, 21], [22, 26], [27, 29], [30, 38], [39, 52], [53, 55], [56, 60], [61, 63], [64, 71], [71, 72], [73, 75], [76, 78], [79, 83], [84, 93], [94, 97], [98, 101], [102, 107], [108, 110], [111, 119], [120, 124], [125, 129], [130, 138], [139, 141], [142, 152], [153, 160], [161, 168], [169, 170], [171, 173], [173, 175], [176, 179], [180, 190], [191, 196], [197, 199], [200, 208], [209, 210], [211, 213], [214, 217], [217, 220], [221, 227], [228, 236], [237, 245], [246, 247], [247, 256], [257, 259], [260, 264], [265, 269], [270, 278], [279, 281], [282, 286], [287, 289], [289, 295], [296, 303], [304, 310], [310, 311], [312, 314], [314, 317], [318, 322], [323, 333], [334, 337], [338, 340], [341, 348], [349, 352], [353, 356], [357, 364], [364, 365]]}
{"doc_key": "ai-train-90", "ner": [[0, 0, "organisation"], [6, 7, "researcher"], [3, 4, "organisation"], [16, 16, "researcher"], [27, 32, "product"], [49, 49, "researcher"], [44, 47, "programlang"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6], "relations": [[0, 0, 3, 4, "related-to", "purchases", false, false], [6, 7, 16, 16, "named", "same", false, false], [6, 7, 49, 49, "named", "same", false, false], [3, 4, 6, 7, "origin", "founded_by", false, false], [27, 32, 0, 0, "artifact", "", false, false], [44, 47, 49, 49, "artifact", "", true, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5], "sentence": ["Unimation", "a", "achet\u00e9", "Vicarm", "Inc.", "de", "Victor", "Scheinman", "en", "1977", ",", "et", "avec", "l'", "aide", "de", "Scheinman", ",", "la", "soci\u00e9t\u00e9", "a", "cr\u00e9\u00e9", "et", "commenc\u00e9", "\u00e0", "produire", "la", "Machine", "Universelle", "Programmable", "pour", "l'", "Assemblage", ",", "un", "nouveau", "mod\u00e8le", "de", "bras", "robotique", ",", "en", "utilisant", "le", "langage", "de", "programmation", "VAL", "de", "Scheinman", "."], "sentence-detokenized": "Unimation a achet\u00e9 Vicarm Inc. de Victor Scheinman en 1977, et avec l'aide de Scheinman, la soci\u00e9t\u00e9 a cr\u00e9\u00e9 et commenc\u00e9 \u00e0 produire la Machine Universelle Programmable pour l'Assemblage, un nouveau mod\u00e8le de bras robotique, en utilisant le langage de programmation VAL de Scheinman.", "token2charspan": [[0, 9], [10, 11], [12, 18], [19, 25], [26, 30], [31, 33], [34, 40], [41, 50], [51, 53], [54, 58], [58, 59], [60, 62], [63, 67], [68, 70], [70, 74], [75, 77], [78, 87], [87, 88], [89, 91], [92, 99], [100, 101], [102, 106], [107, 109], [110, 118], [119, 120], [121, 129], [130, 132], [133, 140], [141, 152], [153, 165], [166, 170], [171, 173], [173, 183], [183, 184], [185, 187], [188, 195], [196, 202], [203, 205], [206, 210], [211, 220], [220, 221], [222, 224], [225, 234], [235, 237], [238, 245], [246, 248], [249, 262], [263, 266], [267, 269], [270, 279], [279, 280]]}
{"doc_key": "ai-train-91", "ner": [[0, 0, "product"], [4, 4, "programlang"], [9, 10, "algorithm"], [13, 18, "product"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[0, 0, 4, 4, "general-affiliation", "", false, false], [0, 0, 9, 10, "origin", "implementation_of", false, false], [0, 0, 13, 18, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2], "sentence": ["J48", "est", "une", "impl\u00e9mentation", "Java", "open", "source", "de", "l'", "algorithme", "C4.5", "dans", "l'", "outil", "d'", "exploration", "de", "donn\u00e9es", "Weka", "."], "sentence-detokenized": "J48 est une impl\u00e9mentation Java open source de l'algorithme C4.5 dans l'outil d'exploration de donn\u00e9es Weka.", "token2charspan": [[0, 3], [4, 7], [8, 11], [12, 26], [27, 31], [32, 36], [37, 43], [44, 46], [47, 49], [49, 59], [60, 64], [65, 69], [70, 72], [72, 77], [78, 80], [80, 91], [92, 94], [95, 102], [103, 107], [107, 108]]}
{"doc_key": "ai-train-92", "ner": [[2, 3, "metrics"], [13, 14, "product"], [21, 28, "misc"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[2, 3, 13, 14, "win-defeat", "", false, false]], "relations_mapping_to_source": [0], "sentence": ["L'", "article", "SSIM", "de", "2004", "a", "\u00e9t\u00e9", "cit\u00e9", "plus", "de", "20 000", "fois", "selon", "Google", "Scholar", ".", "Il", "a", "\u00e9galement", "re\u00e7u", "le", "prix", "IEEE", "Signal", "Processing", "Society", "Sustained", "Impact", "Award", "pour", "2016", ",", "qui", "indique", "qu'", "un", "article", "a", "un", "impact", "inhabituellement", "\u00e9lev\u00e9", "pendant", "au", "moins", "10", "ans", "apr\u00e8s", "sa", "publication", "."], "sentence-detokenized": "L'article SSIM de 2004 a \u00e9t\u00e9 cit\u00e9 plus de 20 000 fois selon Google Scholar. Il a \u00e9galement re\u00e7u le prix IEEE Signal Processing Society Sustained Impact Award pour 2016, qui indique qu'un article a un impact inhabituellement \u00e9lev\u00e9 pendant au moins 10 ans apr\u00e8s sa publication.", "token2charspan": [[0, 2], [2, 9], [10, 14], [15, 17], [18, 22], [23, 24], [25, 28], [29, 33], [34, 38], [39, 41], [42, 48], [49, 53], [54, 59], [60, 66], [67, 74], [74, 75], [76, 78], [79, 80], [81, 90], [91, 95], [96, 98], [99, 103], [104, 108], [109, 115], [116, 126], [127, 134], [135, 144], [145, 151], [152, 157], [158, 162], [163, 167], [167, 168], [169, 172], [173, 180], [181, 184], [184, 186], [187, 194], [195, 196], [197, 199], [200, 206], [207, 223], [224, 229], [230, 237], [238, 240], [241, 246], [247, 249], [250, 253], [254, 259], [260, 262], [263, 274], [274, 275]]}
{"doc_key": "ai-train-93", "ner": [[1, 2, "task"], [33, 34, "product"], [44, 46, "product"], [50, 50, "organisation"], [51, 51, "product"], [56, 56, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5], "relations": [[1, 2, 50, 50, "artifact", "", false, false], [33, 34, 1, 2, "related-to", "performs", false, false], [33, 34, 44, 46, "part-of", "", false, false], [50, 50, 56, 56, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["La", "synth\u00e8se", "vocale", "est", "sur", "le", "point", "d'", "\u00eatre", "totalement", "indiscernable", "de", "la", "voix", "d'", "un", "v\u00e9ritable", "\u00eatre", "humain", "avec", "l'", "introduction", "en", "2016", "du", "logiciel", "d'", "\u00e9dition", "et", "de", "g\u00e9n\u00e9ration", "de", "voix", "Adobe", "Voco", ",", "un", "prototype", "destin\u00e9", "\u00e0", "faire", "partie", "de", "la", "suite", "Adobe", "Creative", ",", "et", "de", "DeepMind", "WaveNet", ",", "un", "prototype", "de", "Google", "."], "sentence-detokenized": "La synth\u00e8se vocale est sur le point d'\u00eatre totalement indiscernable de la voix d'un v\u00e9ritable \u00eatre humain avec l'introduction en 2016 du logiciel d'\u00e9dition et de g\u00e9n\u00e9ration de voix Adobe Voco, un prototype destin\u00e9 \u00e0 faire partie de la suite Adobe Creative, et de DeepMind WaveNet, un prototype de Google.", "token2charspan": [[0, 2], [3, 11], [12, 18], [19, 22], [23, 26], [27, 29], [30, 35], [36, 38], [38, 42], [43, 53], [54, 67], [68, 70], [71, 73], [74, 78], [79, 81], [81, 83], [84, 93], [94, 98], [99, 105], [106, 110], [111, 113], [113, 125], [126, 128], [129, 133], [134, 136], [137, 145], [146, 148], [148, 155], [156, 158], [159, 161], [162, 172], [173, 175], [176, 180], [181, 186], [187, 191], [191, 192], [193, 195], [196, 205], [206, 213], [214, 215], [216, 221], [222, 228], [229, 231], [232, 234], [235, 240], [241, 246], [247, 255], [255, 256], [257, 259], [260, 262], [263, 271], [272, 279], [279, 280], [281, 283], [284, 293], [294, 296], [297, 303], [303, 304]]}
{"doc_key": "ai-train-94", "ner": [[0, 1, "researcher"], [6, 10, "organisation"], [15, 21, "organisation"], [27, 27, "conference"], [32, 36, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[0, 1, 6, 10, "role", "", false, false], [0, 1, 15, 21, "role", "", false, false], [0, 1, 27, 27, "role", "", false, false], [0, 1, 32, 36, "role", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["M.", "Poggio", "est", "membre", "honoraire", "du", "programme", "de", "recherche", "en", "neurosciences", ",", "membre", "de", "l'", "Acad\u00e9mie", "am\u00e9ricaine", "des", "arts", "et", "des", "sciences", ",", "membre", "fondateur", "de", "l'", "AAAI", "et", "membre", "fondateur", "du", "McGovern", "Institute", "for", "Brain", "Research", "."], "sentence-detokenized": "M. Poggio est membre honoraire du programme de recherche en neurosciences, membre de l'Acad\u00e9mie am\u00e9ricaine des arts et des sciences, membre fondateur de l'AAAI et membre fondateur du McGovern Institute for Brain Research.", "token2charspan": [[0, 2], [3, 9], [10, 13], [14, 20], [21, 30], [31, 33], [34, 43], [44, 46], [47, 56], [57, 59], [60, 73], [73, 74], [75, 81], [82, 84], [85, 87], [87, 95], [96, 106], [107, 110], [111, 115], [116, 118], [119, 122], [123, 131], [131, 132], [133, 139], [140, 149], [150, 152], [153, 155], [155, 159], [160, 162], [163, 169], [170, 179], [180, 182], [183, 191], [192, 201], [202, 205], [206, 211], [212, 220], [220, 221]]}
{"doc_key": "ai-train-95", "ner": [[11, 16, "task"], [15, 15, "task"], [22, 23, "task"], [31, 31, "misc"], [30, 32, "misc"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[11, 16, 22, 23, "cause-effect", "", false, false], [15, 15, 22, 23, "cause-effect", "", false, false], [30, 32, 22, 23, "topic", "", false, false], [30, 32, 31, 31, "general-affiliation", "nationality", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Dans", "les", "ann\u00e9es", "90", ",", "encourag\u00e9es", "par", "les", "succ\u00e8s", "de", "la", "reconnaissance", "et", "de", "la", "synth\u00e8se", "vocales", ",", "les", "recherches", "sur", "la", "traduction", "vocale", "ont", "d\u00e9but\u00e9", "avec", "le", "d\u00e9veloppement", "du", "projet", "allemand", "Verbmobil", "."], "sentence-detokenized": "Dans les ann\u00e9es 90, encourag\u00e9es par les succ\u00e8s de la reconnaissance et de la synth\u00e8se vocales, les recherches sur la traduction vocale ont d\u00e9but\u00e9 avec le d\u00e9veloppement du projet allemand Verbmobil.", "token2charspan": [[0, 4], [5, 8], [9, 15], [16, 18], [18, 19], [20, 31], [32, 35], [36, 39], [40, 46], [47, 49], [50, 52], [53, 67], [68, 70], [71, 73], [74, 76], [77, 85], [86, 93], [93, 94], [95, 98], [99, 109], [110, 113], [114, 116], [117, 127], [128, 134], [135, 138], [139, 145], [146, 150], [151, 153], [154, 167], [168, 170], [171, 177], [178, 186], [187, 196], [196, 197]]}
{"doc_key": "ai-train-96", "ner": [[3, 4, "researcher"], [8, 9, "researcher"], [11, 13, "researcher"], [16, 18, "algorithm"], [22, 24, "algorithm"], [29, 29, "algorithm"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5], "relations": [[3, 4, 8, 9, "role", "", false, false], [16, 18, 3, 4, "origin", "", false, false], [16, 18, 8, 9, "origin", "", false, false], [16, 18, 11, 13, "origin", "", false, false], [16, 18, 29, 29, "part-of", "", false, false], [22, 24, 16, 18, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5], "sentence": ["En", "1999", ",", "Felix", "Gers", "et", "son", "conseiller", "J\u00fcrgen", "Schmidhuber", "et", "Fred", "Cummins", "ont", "introduit", "la", "porte", "d'", "oubli", "(", "\u00e9galement", "appel\u00e9e", "porte", "de", "maintien", ")", "dans", "l'", "architecture", "LSTM", ","], "sentence-detokenized": "En 1999, Felix Gers et son conseiller J\u00fcrgen Schmidhuber et Fred Cummins ont introduit la porte d'oubli (\u00e9galement appel\u00e9e porte de maintien) dans l'architecture LSTM,", "token2charspan": [[0, 2], [3, 7], [7, 8], [9, 14], [15, 19], [20, 22], [23, 26], [27, 37], [38, 44], [45, 56], [57, 59], [60, 64], [65, 72], [73, 76], [77, 86], [87, 89], [90, 95], [96, 98], [98, 103], [104, 105], [105, 114], [115, 122], [123, 128], [129, 131], [132, 140], [140, 141], [142, 146], [147, 149], [149, 161], [162, 166], [166, 167]]}
{"doc_key": "ai-train-97", "ner": [[2, 5, "field"], [8, 11, "field"], [14, 16, "algorithm"]], "ner_mapping_to_source": [0, 1, 2], "relations": [[14, 16, 2, 5, "part-of", "", false, false], [14, 16, 8, 11, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1], "sentence": ["Dans", "le", "traitement", "du", "signal", "num\u00e9rique", "et", "la", "th\u00e9orie", "de", "l'", "information", ",", "la", "fonction", "sinc", "normalis\u00e9e", "est", "g\u00e9n\u00e9ralement", "d\u00e9finie", "comme", "suit"], "sentence-detokenized": "Dans le traitement du signal num\u00e9rique et la th\u00e9orie de l'information, la fonction sinc normalis\u00e9e est g\u00e9n\u00e9ralement d\u00e9finie comme suit", "token2charspan": [[0, 4], [5, 7], [8, 18], [19, 21], [22, 28], [29, 38], [39, 41], [42, 44], [45, 52], [53, 55], [56, 58], [58, 69], [69, 70], [71, 73], [74, 82], [83, 87], [88, 98], [99, 102], [103, 115], [116, 123], [124, 129], [130, 134]]}
{"doc_key": "ai-train-98", "ner": [[3, 8, "field"], [10, 11, "researcher"], [17, 20, "conference"], [24, 28, "organisation"], [30, 30, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3, 4], "relations": [[3, 8, 10, 11, "origin", "coined_term", false, false], [10, 11, 17, 20, "role", "", false, false], [10, 11, 24, 28, "role", "", false, false], [30, 30, 24, 28, "named", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3], "sentence": ["Le", "terme", "\"", "linguistique", "informatique", "\"", "a", "\u00e9t\u00e9", "invent\u00e9", "par", "David", "Hays", ",", "membre", "fondateur", "de", "l'", "Association", "for", "Computational", "Linguistics", "et", "de", "l'", "International", "Committee", "on", "Computational", "Linguistics", "(", "ICCL", ")", "."], "sentence-detokenized": "Le terme \"linguistique informatique\" a \u00e9t\u00e9 invent\u00e9 par David Hays, membre fondateur de l'Association for Computational Linguistics et de l'International Committee on Computational Linguistics (ICCL).", "token2charspan": [[0, 2], [3, 8], [9, 10], [10, 22], [23, 35], [35, 36], [37, 38], [39, 42], [43, 50], [51, 54], [55, 60], [61, 65], [65, 66], [67, 73], [74, 83], [84, 86], [87, 89], [89, 100], [101, 104], [105, 118], [119, 130], [131, 133], [134, 136], [137, 139], [139, 152], [153, 162], [163, 165], [166, 179], [180, 191], [192, 193], [193, 197], [197, 198], [198, 199]]}
{"doc_key": "ai-train-99", "ner": [[10, 21, "misc"], [9, 9, "misc"], [37, 39, "metrics"], [41, 41, "metrics"]], "ner_mapping_to_source": [0, 1, 2, 3], "relations": [[41, 41, 37, 39, "named", "", false, false]], "relations_mapping_to_source": [0], "sentence": ["59", ",", "pp.", "2547-2553", ",", "oct.", "2011", "Dans", "le", "DPD", "\u00e0", "m\u00e9moire", "(", "ou", "sans", "m\u00e9moire", ")", "\u00e0", "base", "de", "polyn\u00f4mes", "unidimensionnels", ",", "afin", "de", "r\u00e9soudre", "les", "coefficients", "des", "polyn\u00f4mes", "du", "pr\u00e9-distueur", "num\u00e9rique", "et", "de", "minimiser", "l'", "erreur", "quadratique", "moyenne", "(", "EQM", ")", ",", "la", "sortie", "d\u00e9form\u00e9e", "du", "syst\u00e8me", "non", "lin\u00e9aire", "doit", "\u00eatre", "sur-\u00e9chantillonn\u00e9e", "\u00e0", "un", "taux", "qui", "permet", "la", "capture", "des", "produits", "non", "lin\u00e9aires", "de", "l'", "ordre", "du", "pr\u00e9-distueur", "num\u00e9rique", "."], "sentence-detokenized": "59, pp. 2547-2553, oct. 2011 Dans le DPD \u00e0 m\u00e9moire (ou sans m\u00e9moire) \u00e0 base de polyn\u00f4mes unidimensionnels, afin de r\u00e9soudre les coefficients des polyn\u00f4mes du pr\u00e9-distueur num\u00e9rique et de minimiser l'erreur quadratique moyenne (EQM), la sortie d\u00e9form\u00e9e du syst\u00e8me non lin\u00e9aire doit \u00eatre sur-\u00e9chantillonn\u00e9e \u00e0 un taux qui permet la capture des produits non lin\u00e9aires de l'ordre du pr\u00e9-distueur num\u00e9rique.", "token2charspan": [[0, 2], [2, 3], [4, 7], [8, 17], [17, 18], [19, 23], [24, 28], [29, 33], [34, 36], [37, 40], [41, 42], [43, 50], [51, 52], [52, 54], [55, 59], [60, 67], [67, 68], [69, 70], [71, 75], [76, 78], [79, 88], [89, 105], [105, 106], [107, 111], [112, 114], [115, 123], [124, 127], [128, 140], [141, 144], [145, 154], [155, 157], [158, 170], [171, 180], [181, 183], [184, 186], [187, 196], [197, 199], [199, 205], [206, 217], [218, 225], [226, 227], [227, 230], [230, 231], [231, 232], [233, 235], [236, 242], [243, 251], [252, 254], [255, 262], [263, 266], [267, 275], [276, 280], [281, 285], [286, 304], [305, 306], [307, 309], [310, 314], [315, 318], [319, 325], [326, 328], [329, 336], [337, 340], [341, 349], [350, 353], [354, 363], [364, 366], [367, 369], [369, 374], [375, 377], [378, 390], [391, 400], [400, 401]]}
{"doc_key": "ai-train-100", "ner": [[0, 1, "researcher"], [10, 10, "location"], [12, 14, "location"], [16, 17, "country"], [21, 21, "location"], [23, 23, "country"], [35, 43, "organisation"], [45, 48, "organisation"], [50, 50, "location"], [54, 55, "organisation"]], "ner_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], "relations": [[0, 1, 10, 10, "physical", "", false, false], [0, 1, 45, 48, "physical", "", false, false], [0, 1, 54, 55, "role", "", false, false], [10, 10, 12, 14, "physical", "", false, false], [12, 14, 16, 17, "physical", "", false, false], [35, 43, 45, 48, "part-of", "", false, false], [45, 48, 50, 50, "physical", "", false, false], [54, 55, 35, 43, "part-of", "", false, false]], "relations_mapping_to_source": [0, 1, 2, 3, 4, 5, 6, 7], "sentence": ["Boris", "Katz", ",", "(", "n\u00e9", "le", "5", "octobre", "1947", "\u00e0", "Chi\u0219in\u0103u", ",", "RSS", "de", "Moldavie", ",", "Union", "sovi\u00e9tique", ",", "(", "aujourd'hui", "Chi\u0219in\u0103u", ",", "Moldavie", ")", ")", "est", "un", "chercheur", "principal", "am\u00e9ricain", "(", "informaticien", ")", "au", "laboratoire", "d'", "informatique", "et", "d'", "intelligence", "artificielle", "du", "MIT", "au", "Massachusetts", "Institute", "of", "Technology", "\u00e0", "Cambridge", "et", "chef", "du", "groupe", "InfoLab", "du", "laboratoire", "."], "sentence-detokenized": "Boris Katz, (n\u00e9 le 5 octobre 1947 \u00e0 Chi\u0219in\u0103u, RSS de Moldavie, Union sovi\u00e9tique, (aujourd'hui Chi\u0219in\u0103u, Moldavie)) est un chercheur principal am\u00e9ricain (informaticien) au laboratoire d'informatique et d'intelligence artificielle du MIT au Massachusetts Institute of Technology \u00e0 Cambridge et chef du groupe InfoLab du laboratoire.", "token2charspan": [[0, 5], [6, 10], [10, 11], [12, 13], [13, 15], [16, 18], [19, 20], [21, 28], [29, 33], [34, 35], [36, 44], [44, 45], [46, 49], [50, 52], [53, 61], [61, 62], [63, 68], [69, 79], [79, 80], [81, 82], [82, 93], [94, 102], [102, 103], [104, 112], [112, 113], [113, 114], [115, 118], [119, 121], [122, 131], [132, 141], [142, 151], [152, 153], [153, 166], [166, 167], [168, 170], [171, 182], [183, 185], [185, 197], [198, 200], [201, 203], [203, 215], [216, 228], [229, 231], [232, 235], [236, 238], [239, 252], [253, 262], [263, 265], [266, 276], [277, 278], [279, 288], [289, 291], [292, 296], [297, 299], [300, 306], [307, 314], [315, 317], [318, 329], [329, 330]]}