| { |
| "paper_id": "W17-0214", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T04:30:06.010913Z" |
| }, |
| "title": "North S\u00e1mi to Finnish rule-based machine translation system", |
| "authors": [ |
| { |
| "first": "Ryan", |
| "middle": [], |
| "last": "Johnson\u00b9", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Giela ja kultuvrra instituhtta", |
| "location": { |
| "settlement": "Romssa" |
| } |
| }, |
| "email": "ryan.txanson@gmail.com" |
| }, |
| { |
| "first": "Tommi", |
| "middle": [ |
| "A" |
| ], |
| "last": "Pirinen\u00b2", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "tommi.antero.pirinen@uni-hamburg.de" |
| }, |
| { |
| "first": "Tiina", |
| "middle": [], |
| "last": "Puolakainen\u00b3", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "tiina.puolakainen@eki.ee" |
| }, |
| { |
| "first": "Francis", |
| "middle": [], |
| "last": "Tyers\u00b9", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Giela ja kultuvrra instituhtta", |
| "location": { |
| "settlement": "Romssa" |
| } |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Trond", |
| "middle": [], |
| "last": "Trosterud\u00b9", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Giela ja kultuvrra instituhtta", |
| "location": { |
| "settlement": "Romssa" |
| } |
| }, |
| "email": "trond.trosterud@uit.no" |
| }, |
| { |
| "first": "Kevin", |
| "middle": [], |
| "last": "Unhammer\u00b9", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Giela ja kultuvrra instituhtta", |
| "location": { |
| "settlement": "Romssa" |
| } |
| }, |
| "email": "" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "This paper presents a machine translation system between Finnish and North S\u00e1mi, two Uralic languages. In this paper we concentrate on the translation direction to Finnish. As a background, the differences between the two languages is presented, followed by how the system was designed to handle some of these differences. We then provide an evaluation of the system's performance and directions for future work.", |
| "pdf_parse": { |
| "paper_id": "W17-0214", |
| "_pdf_hash": "", |
| "abstract": [ |
| { |
| "text": "This paper presents a machine translation system between Finnish and North S\u00e1mi, two Uralic languages. In this paper we concentrate on the translation direction to Finnish. As a background, the differences between the two languages is presented, followed by how the system was designed to handle some of these differences. We then provide an evaluation of the system's performance and directions for future work.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Abstract", |
| "sec_num": null |
| } |
| ], |
| "body_text": [ |
| { |
| "text": "This paper presents a prototype shallow-transfer rule-based machine translation system between Finnish and North S\u00e1mi. The paper will be laid out as follows: Section 2 gives a short review of some previous work in the area of Uralic-Uralic language machine translation; Section 3 introduces Finnish and North S\u00e1mi and compares their grammar; Section 4 describes the system and the tools used to construct it; Section 5 gives a preliminary evaluation of the system; and finally Section 6 describes our aims for future work and some concluding remarks.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction 0", |
| "sec_num": "1" |
| }, |
| { |
| "text": "Within the Apertium platform, work on several MT systems from North S\u00e1mi to Norwegian and to other S\u00e1mi languages have been developed (Tyers et al., 2009; Wiechetek et al., 2010; Trosterud and Unhammer, 2013; Antonsen et al., 2016) ). Besides these systems, several previous works on making machine translation systems between Uralic languages exist, although to our knowledge none are publicly available, except for North S\u00e1mi 0 Authors are listed here alphabetically to Norwegian 1 , and the translation between Estonian, Finnish and Hungarian being available via English as a pivot language in Google Translate. 2 For non-Uralic pairs there are also numerous similarly laid out systems e.g. in Apertium's Turkic pairs, e.g. (Salimzyanov et al., 2013) , that can offer insights on how the pair is implemented, which are detailed later in the article but the main parts are the same.", |
| "cite_spans": [ |
| { |
| "start": 134, |
| "end": 154, |
| "text": "(Tyers et al., 2009;", |
| "ref_id": "BIBREF17" |
| }, |
| { |
| "start": 155, |
| "end": 178, |
| "text": "Wiechetek et al., 2010;", |
| "ref_id": "BIBREF18" |
| }, |
| { |
| "start": 179, |
| "end": 208, |
| "text": "Trosterud and Unhammer, 2013;", |
| "ref_id": "BIBREF15" |
| }, |
| { |
| "start": 209, |
| "end": 231, |
| "text": "Antonsen et al., 2016)", |
| "ref_id": "BIBREF1" |
| }, |
| { |
| "start": 615, |
| "end": 616, |
| "text": "2", |
| "ref_id": null |
| }, |
| { |
| "start": 727, |
| "end": 753, |
| "text": "(Salimzyanov et al., 2013)", |
| "ref_id": "BIBREF12" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Previous work", |
| "sec_num": "2" |
| }, |
| { |
| "text": "North S\u00e1mi and Finnish belong to the S\u00e1mi and Finnic branches of the Uralic languages, respectively. The languages are mutually unintelligible, but grammatically quite similar. The orthographical conventions between Finnish and North S\u00e1mi written in Finland were quite similar until 1979, when an unified North S\u00e1mi orthography widened the distance to Finnish. Finnish is primarily spoken in Finland, where it is the national language, sharing status with Swedish as an official language. The total number of speakers is at least 6 million people. North S\u00e1mi is spoken in the Northern parts of Norway, Sweden and Finland by approximately 24.700 people, and it has, alongside the national language, some official status in the municipalities and counties where it is spoken. North S\u00e1mi speakers are bilingual in their mother tongue and in their respective national language, many also speak the neighbouring official language. An MT system between North S\u00e1mi and Finnish is potentially of great use to the language communities, although fulfilling different functions. In Finland, it may be used to understand S\u00e1mi text, and in Norway and Sweden, it may be used by North S\u00e1mi speakers to understand Finnish text. In principle, the system may also be used for North S\u00e1mi text production, although further de-velopment will be needed to fulfil such a function.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The languages", |
| "sec_num": "3" |
| }, |
| { |
| "text": "As related languages, Finnish and North S\u00e1mi share several phonological processes, the most important one being consonant gradation. However, North S\u00e1mi consonant gradation involves the vast majority of stem-internal consonant clusters, whereas the Finnish counterpart involves only the stops p, t, k. Vowel length has a more central role in Finnish than in North S\u00e1mi, Several instances of final vowel apocopy in North S\u00e1mi, as well as a neutralisation of p, t, k in word-final position, has also resulted in quite extensive morphological homonymy. A richer inventory of affricates and fricatives in North S\u00e1mi, as well as preaspiration, also add to the difference.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Phonological differences", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "In the native vocabulary, neither Finnish nor North S\u00e1mi distinguish between voiced and unvoiced plosives, but whereas Finnish writes them as p, t, k, North S\u00e1mi writes b, d, g, as in kirja : girji \"book\". Finnish marks vowel length with double letter symbols. In North S\u00e1mi this distinction is marked for one vowel only, a, and with acute accent. Apart from this the orthographic principles of the two languages is quite similar, the almost total lack of free rides is a result of different phonology.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Orthographic differences", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "There are a number of examples where the morphologies of Finnish and North S\u00e1mi are rather different.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological differences", |
| "sec_num": "3.3" |
| }, |
| { |
| "text": "North S\u00e1mi has a separate dual number, whereas Finnish has not. Otherwise the North S\u00e1mi and Finnish finite verb morphology is almost identical. The infinite verb conjugation is more different, though: Finnish has a rich array of infinitives that are inflected in different subsets of the case system.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological differences", |
| "sec_num": "3.3" |
| }, |
| { |
| "text": "Finnish has more than twice the number of cases as North S\u00e1mi has. Where North S\u00e1mi only has one case for the direct object (accusative), Finnish has two (accusative and partitive). The Finnish system of adverbial cases consist of a 2x3 matrix of inner/outer to/in/from cases, North S\u00e1mi has only one of these distinctions (to/in\u02dcfrom), thus the 6 Finnish cases corresponds to 2 North S\u00e1mi ones. In principle, Finnish and North S\u00e1mi have the same system of possessive suffixes, but in North S\u00e1mi its use is far more restricted than in Finnish.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological differences", |
| "sec_num": "3.3" |
| }, |
| { |
| "text": "Syntactically speaking, there are two varieties of North S\u00e1mi, one used within and one outside of Finland. The Finnish variety is much closer to Finnish than the Scandinavian one. Comparing Finnish with the Scandinavian variety of North S\u00e1mi, the most striking difference is participle constructions vs. relative clauses. Where North S\u00e1mi uses subordinate clauses, written Finnish often use head-final participle constructions instead. Since both varieties are found in Finnish, at least to some degree, we at the moment let most \"Scandinavian\" varieties of North S\u00e1mi through, thereby giving North S\u00e1mi from Norway and Finland a different stylistic flavour in the Finnish output.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Syntactic differences", |
| "sec_num": "3.4" |
| }, |
| { |
| "text": "The North S\u00e1mi passive is a derivational process, whereas it for Finnish is an inflectional one, resulting in quite different syntactic patterns for passive. Finnish has a richer array of indefinite verb forms.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Syntactic differences", |
| "sec_num": "3.4" |
| }, |
| { |
| "text": "Finnish adjectives agree with their head noun in case and number, whereas North S\u00e1mi has an invariant attribute form for all but one adjective, the adjective buorre 'good', and partial agreement for determiners.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Syntactic differences", |
| "sec_num": "3.4" |
| }, |
| { |
| "text": "Existential and habitive clauses have the same structure in the two languages, possessor.localcase copula possessed and adverbial copula esubject (on me / in street is car 'I have a car/There is a car in the street'). except that in Finnish, the possessed/e-subject behaves like objects, whereas it in North S\u00e1mi they behave like subjects. Thus, in North S\u00e1mi, the copula agrees with the possessed / e-subject, whereas in Finnish, it does not.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Syntactic differences", |
| "sec_num": "3.4" |
| }, |
| { |
| "text": "The Apertium translation engine consists of a Unix-style pipeline or assembly line with the following modules (see Figure 1 ):", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 115, |
| "end": 123, |
| "text": "Figure 1", |
| "ref_id": "FIGREF0" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Architecture of the system", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "\u2022 A deformatter which encapsulates the format information in the input as superblanks that will then be seen as blanks between words by the other modules.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Architecture of the system", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "\u2022 A morphological analyser which segments the text in surface forms (SF) (words, or, where detected, multiword lexical units or MWLUs) and for each, delivers one or more lexical forms (LF) consisting of lemma, lexical category and morphological information.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Architecture of the system", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "\u2022 A morphological disambiguator (CG) which chooses, using linguistic rules the most adequate sequence of morphological analyses for an ambiguous sentence.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Architecture of the system", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "\u2022 A lexical transfer module which reads each SL LF and delivers the corresponding targetlanguage (TL) LF by looking it up in a bilingual dictionary encoded as an FST compiled from the corresponding XML file. The lexical transfer module may return more than one TL LF for a single SL LF.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Architecture of the system", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "\u2022 A lexical selection module (Tyers et al., 2012b) which chooses, based on context rules, the most adequate translation of ambiguous source language LFs.", |
| "cite_spans": [ |
| { |
| "start": 29, |
| "end": 50, |
| "text": "(Tyers et al., 2012b)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Architecture of the system", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "\u2022 A structural transfer module, which performs local syntactic operations, is compiled from XML files containing rules that associate an action to each defined LF pattern. Patterns are applied left-to-right, and the longest matching pattern is always selected.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Architecture of the system", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "\u2022 A morphological generator which delivers a TL SF for each TL LF, by suitably inflecting it.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Architecture of the system", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "\u2022 A reformatter which de-encapsulates any format information. Table 1 provides an example of a single phrase as it moves through the pipeline.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 62, |
| "end": 69, |
| "text": "Table 1", |
| "ref_id": "TABREF0" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Architecture of the system", |
| "sec_num": "4.1" |
| }, |
| { |
| "text": "The morphological transducers are compiled with the Helsinki Finite State Technology (Lind\u00e9n et al., 2009) , 5 a free/open-source reimplementation of the Xerox finite-state tool-chain, popular in the field of morphological analysis. It implements both the lexc morphology description language for defining lexicons, and the twol and xfst scripting languages for modeling morphophonological rules. This toolkit has been chosen as it-or the equivalent XFST-has been widely used for other Uralic languages (Koskenniemi, 1983; Pirinen, 2015; Moshagen et al., 2013) , and is available under a free/open-source licence. The morphologies of both languages are implemented in lexc, and the morphophonologies of both languages are implemented in twolc.", |
| "cite_spans": [ |
| { |
| "start": 85, |
| "end": 106, |
| "text": "(Lind\u00e9n et al., 2009)", |
| "ref_id": "BIBREF9" |
| }, |
| { |
| "start": 109, |
| "end": 110, |
| "text": "5", |
| "ref_id": null |
| }, |
| { |
| "start": 503, |
| "end": 522, |
| "text": "(Koskenniemi, 1983;", |
| "ref_id": "BIBREF7" |
| }, |
| { |
| "start": 523, |
| "end": 537, |
| "text": "Pirinen, 2015;", |
| "ref_id": "BIBREF11" |
| }, |
| { |
| "start": 538, |
| "end": 560, |
| "text": "Moshagen et al., 2013)", |
| "ref_id": "BIBREF10" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological transducers", |
| "sec_num": "4.2" |
| }, |
| { |
| "text": "The same morphological description is used for both analysis and generation. To avoid overgeneration, any alternative forms are marked with one of two marks, LR (only analyser) or RL (only generator). Instead of the usual compile/invert to compile the transducers, we compile twice, once the generator, without the LR paths, and then again the analyser without the RL paths.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological transducers", |
| "sec_num": "4.2" |
| }, |
| { |
| "text": "The bilingual lexicon currently contains 19,415 stem-to-stem correspondences (of which 8044 proper nouns) and was built partly upon an available North S\u00e1mi-Finnish dictionary 6 , and partly by hand (i.e., by translating North S\u00e1mi stems unrecognised by the morphological analyser into Finnish). The proper nouns were taken from existing lexical resources. Entries consist largely of one-to-one stem-to-stem correspondences with part of speech, but also include some entries with ambiguous translations (see e.g., Figure 2 ).", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 513, |
| "end": 521, |
| "text": "Figure 2", |
| "ref_id": "FIGREF1" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Bilingual lexicon", |
| "sec_num": "4.3" |
| }, |
| { |
| "text": "The system has a morphological disambiguation module in the form of a Constraint Grammar (Karlsson et al., 1995) . The version of the formalism used is vislcg3. 7 The output of each morphological analyser is highly ambiguous, measured at around 2.4 morphological analyses per form for Finnish and 2.6 for North S\u00e1mi 8 . The goal of Morphological analysis: S\u00e1megielat/ s\u00e1megielat<adj> <attr>/ s\u00e1megielat<adj> <sg><nom>/ s\u00e1megiella<n> <pl><nom>/ s\u00e1megiella<n> <sg><acc><px2sg>/ s\u00e1megiella<n> <sg><gen><px2sg>/ s\u00e1megiella<n> <sg><acc><px2sg>/ s\u00e1megiella<n> <sg><gen><px2sg>$ leat/ leat<vblex><iv> <indic><pres><conneg>/ leat<vblex><iv> <indic><pres><p1><pl>/ leat<vblex><iv> <indic><pres><p2><sg>/ leat<vblex><iv> <indic><pres><p3><pl>/ leat<vblex><iv> <inf>$ gielat/ giella<n> <pl><nom>/ giella<n> <sg><acc><px2sg>/ giella<n> <sg><gen><px2sg>/ giella<n> <sg><acc><px2sg>/ giella<n> <sg><gen><px2sg>$\u02c6maid/ maid<adv>/ mii<prn><itg> <pl><acc>/ mii<prn><itg> <pl><gen>/ mii<prn><itg> <sg><acc>/ mii<prn><rel> <pl><acc>/ mii<prn><rel> <pl><gen>/ mii<prn><rel> <sg><acc>$ s\u00e1mit/ s\u00e1pmi<n> <pl><nom>/ s\u00e1pmi<n> <pl><nom>$ h\u00e1llet/ h\u00e1llat<vblex><tv> <imp><p2><pl>/ h\u00e1llat<vblex><tv> <indic><pres><p3><pl>/ h\u00e1llat<vblex><tv> <indic><pret><p2><sg>$ ./.<sent>$ Morphological disambiguation: S\u00e1megielat/s\u00e1megiella<n> <pl><nom> <@SUBJ\u2192>$ leat/leat<vblex><iv> <indic><pres><p3><pl> <@+FMAINV>$ gielat/giella<n> <pl><nom> <@\u2190SPRED>$ maid/mii<prn><rel> <pl><acc> <@OBJ\u2192>$ s\u00e1mit/s\u00e1pmi<n> <pl><nom> <@SUBJ\u2192>$ h\u00e1llet/h\u00e1llat<vblex><tv> <indic><pres><p3><pl> <@+FMAINV>$\u02c6./.<sent>$ Lexical translation: S\u00e1megiella<n> <pl><nom> <@SUBJ\u2192>/ Saamekieli<n> <pl><nom> <@SUBJ\u2192>/ Saame<n> <pl><nom> <@SUBJ\u2192>$ leat<vblex><iv> <indic><pres><p3><pl> <@+FMAINV>/ olla<vblex> <actv><indic><pres><p3><pl> <@+FMAINV>/ sijaita<vblex> <actv><indic><pres><p3><pl> <@+FMAINV>$ giella<n> <pl><nom> <@\u2190SPRED>/ kieli<n> <pl><nom> <@\u2190SPRED>/ ansa<n> <pl><nom> <@\u2190SPRED>$ mii<prn><rel> <pl><acc> <@OBJ\u2192>/ mik\u00e4<prn><rel> <pl><acc> <@OBJ\u2192>$ s\u00e1pmi<n> <pl><nom> <@SUBJ\u2192>/ saame<n> <pl><nom> <@SUBJ\u2192>$ h\u00e1llat<vblex><tv> <indic><pres><p3><pl> <@+FMAINV>/ puhua<vblex> <actv><indic><pres><p3><pl> <@+FMAINV>/ mekastaa<vblex> <actv><indic><pres><p3><pl> <@+FMAINV>$\u02c6.<sent>/.<sent>$ Structural transfer: Saamekieli<n> <pl><nom>$\u02c6olla<vblex> <actv><indic><pres><p3><pl>$ kieli<n> <pl><nom>$\u02c6mik\u00e4<prn><rel> <pl><par>$ saame<n> <pl><nom>$\u02c6puhua<vblex> <actv><indic><pres><p3><pl>$\u02c6.<sent>$ Finnish translation: Saamekielet ovat kielet #mik\u00e4 saamet puhuvat <e><p><l>s\u00e1lten<s n=\"n\"/></l><r>suolaus<s n=\"n\"/></r></p></e> <e><p><l>s\u00e1lti<s n=\"n\"/></l><r>suola<s n=\"n\"/></r></p></e> <e><p><l>s\u00e1meduodji<s n=\"n\"/></l><r>k\u00e4sity\u00f6<s n=\"n\"/></r></p></e> <e><p><l>s\u00e1megiella<s n=\"n\"/></l><r>saame<s n=\"n\"/></r></p></e> <e><p><l>s\u00e1megiella<s n=\"n\"/></l><r>saamekieli<s n=\"n\"/></r></p></e> <e><p><l>s\u00e1mi<s n=\"n\"/></l><r>saame<s n=\"n\"/></r></p></e> <e><p><l>s\u00e1mil<s n=\"n\"/></l><r>sammal<s n=\"n\"/></r></p></e> the CG rules is to select the correct analysis when there are multiple analyses. Given the similarity of Finnish and North S\u00e1mi, ambiguity across parts of speech may often be passed from one language to the other and not lead to many translation errors. Disambiguating between forms within the inflectional paradigms in case of homonymy, on the other hand, are crucial for choosing the correct form of the target language, and there has been put much effort into developing CG rules to resolve such ambiguity for North S\u00e1mi. Currently, ambiguity is down to 1.08 for North S\u00e1mi (analysed with the disambigator used for MT on a 675534 word newspaper corpus 9 . The corresponding number for Finnish is 1.36, for a subcorpus of 770999 words of Wikipedia text. The Finnish CG rules are a conversion of Fred Karlsson's original CG1 rules for Finnish (Karlsson, 1990) , and the poorer results for Finnish are due to conversion problems between the different CG version, and between CG1 and our Finnish FST.", |
| "cite_spans": [ |
| { |
| "start": 89, |
| "end": 112, |
| "text": "(Karlsson et al., 1995)", |
| "ref_id": "BIBREF5" |
| }, |
| { |
| "start": 161, |
| "end": 162, |
| "text": "7", |
| "ref_id": null |
| }, |
| { |
| "start": 3777, |
| "end": 3793, |
| "text": "(Karlsson, 1990)", |
| "ref_id": "BIBREF6" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Disambiguation rules", |
| "sec_num": "4.4" |
| }, |
| { |
| "text": "All evaluation was tested against a specific version of Apertium SVN 10 and Giellatekno SVN 11 . The lexical coverage of the system was calculated over freely available corpora of North S\u00e1mi. We used a recent dump of Wikipedia 12 as well as a translation of the New Testament. The corpora were divided into 10 parts each; the coverage numbers given are the averages of the calculated percentages of number of words analysed for each of these parts, and the standard deviation presented is the", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Evaluation", |
| "sec_num": "5" |
| }, |
| { |
| "text": "Tokens Cov. std se.wikipedia.org 190,894 76,81 % \u00b110 New Testament 162,718 92,45 % \u00b10.06 Table 2 : Na\u00efve coverage of sme-fin system standard deviation of the coverage on each corpus. As shown in Table 2 , the na\u00efve coverage 13 of the North S\u00e1mi to Finnish MT system over the corpora approaches that of a broad-coverage MT system, with one word in ten unknown.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 89, |
| "end": 96, |
| "text": "Table 2", |
| "ref_id": null |
| }, |
| { |
| "start": 195, |
| "end": 202, |
| "text": "Table 2", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "Corpus", |
| "sec_num": null |
| }, |
| { |
| "text": "The coverage over the Wikipedia corpus is substantially worse, due to the fact that this corpus is \"dirtier\": it contains orthographical errors, wiki code 14 , repetitions, lot of English texts, as well as quite a few proper nouns, this is easily seen in the large deviation between divided parts. The New Testament on the other hand is rather well covered and has practically uniformly distributed coverage throughout.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Corpus", |
| "sec_num": null |
| }, |
| { |
| "text": "To measure the performance of the translator we used the Word Error Rate metric-an edit-distance metric based on Levenshtein distance (Levenshtein, 1966) . We had three small North S\u00e1mi corpora along with their manually post-edited translations into Finnish to measure the WER. We have chosen not to measure the translation quality with automatic measures such as BLEU, as they are not the best suited to measure quality of translations for the use case, for further details see also Callison-Burch et al. (2006; Smith et al. (2016; Smith et al. (2014) .", |
| "cite_spans": [ |
| { |
| "start": 134, |
| "end": 153, |
| "text": "(Levenshtein, 1966)", |
| "ref_id": "BIBREF8" |
| }, |
| { |
| "start": 484, |
| "end": 512, |
| "text": "Callison-Burch et al. (2006;", |
| "ref_id": "BIBREF2" |
| }, |
| { |
| "start": 513, |
| "end": 532, |
| "text": "Smith et al. (2016;", |
| "ref_id": "BIBREF14" |
| }, |
| { |
| "start": 533, |
| "end": 552, |
| "text": "Smith et al. (2014)", |
| "ref_id": "BIBREF13" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Corpus", |
| "sec_num": null |
| }, |
| { |
| "text": "For translation post-edition we used three freely 13 A non-na\u00efve coverage would require manual evaluation of correctness for the cases where word-forms are covered accidentally by e.g., morphological processes.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Corpus", |
| "sec_num": null |
| }, |
| { |
| "text": "14 While we have tried to clean the data from most of the Mediawiki codes and notations, there is always some left after the cleanup, due to new wiki codes after creation of the cleanup script or actual broken data available parallel texts from the internet: one from the Finnish S\u00e1mi parliament site 15 , one from a Swedish regulation of minority people and languages and the story that is used with all Apertium language pairs as initial development set (\"Where is James?\"). Table 3 presents the WER for the corpora. Analysing the changes in post-edition, a few classes of actual errors can be identified. One common example arises from the grammatical differences in the case system systems, in particular the remaining adpositions are often turned into a case suffix for the dependent noun phrase, e.g., the North S\u00e1mi \"birra\" has been turned into the Finnish adposition \"ymp\u00e4ri\" (around), where elative case is required, similarly for the translation \"seassa\" (among) instead of inessive case. Also visible, especially in the story text is the lack of possessive suffix agreement e.g. \"heid\u00e4n \u00e4iti\" (their mother n sg nom) instead of \"heid\u00e4n \u00e4itins\u00e4\" (their mother n sp nom/gen pxsp3), while the former is perfectly acceptable in standard spoken Finnish it is not accepted as formal written language form. Another issue that appeared a number of times, maybe partially due to the genre of the texts selected, i.e. law texts, was the selection of adverb (form), e.g. the word-form \"mukana\" (with) was corrected to \"mukaan\" (according to). A large amount of simple lexical problems is due to the vocabulary of the selected texts as well: \"hallintoalue\" (governmental area), \"seurantavastuu\" (responsibility of surveillance), \"itsehallinto\" (autonomy), and their compounds, are all either missing or partially wrong due to lexical selections.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 477, |
| "end": 484, |
| "text": "Table 3", |
| "ref_id": "TABREF2" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Corpus", |
| "sec_num": null |
| }, |
| { |
| "text": "We have presented the first MT system from Finnish to North S\u00e1mi. With a WER of above 30%, it still is far from production-level performance, and it is also at the prototype-level in 15 http://samediggi.fi terms of the number of rules. Although the impact of this relatively low number of rules on the quality of translation is extensive (cf., the difference in WER between the development and testing corpora), the outlook is promising and the current results suggest that a high quality translation between morphologically-rich agglutinative languages is possible. We plan to continue development on the pair; the coverage of the system is already quite high, although we intend to increase it to 95 % on the corpora we have we estimate that this will mean adding around 5,000 new stems and take 1-2 months. The remaining work will be improving the quality of translation by adding more rules, starting with the transfer component. The long-term plan is to integrate the data created with other open-source data for Uralic languages in order to make transfer systems between all the Uralic language pairs. Related work is currently ongoing from North S\u00e1mi to South, Lule and Inari S\u00e1mi, from North S\u00e1mi to Norwegian, and between Finnish and Estonian. The system presented here is available as free/open-source software under the GNU GPL and the whole system may be downloaded from Sourceforge and the open repository of Giellatekno.", |
| "cite_spans": [ |
| { |
| "start": 183, |
| "end": 185, |
| "text": "15", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Concluding remarks", |
| "sec_num": "6" |
| }, |
| { |
| "text": "https://gtweb.uit.no/jorgal 2 https://translate.google.com", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "SystemThe system is based on the Apertium 3 machine translation platform(Forcada et al., 2011). The platform was originally aimed at the Romance languages of the Iberian peninsula, but has also been adapted for other, more distantly related, language pairs. The whole platform, both programs and data, are licensed under the Free Software Foundation's General Public Licence 4 (GPL) and all the software and data for the 30 released language pairs (and the other pairs being worked on) is available for download from the project website.3 http://apertium.sf.net 4 https://www.gnu.org/licenses/gpl-3.0.en. html", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "https://hfst.github.io 6 https://gtweb.uit.no/langtech/trunk/words/ dicts/smefin/src/ 7 http://visl.cg.sdu.dk 8 Cf.(Trosterud and Wiechetek, 2007)", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "Cf.(Antonsen and Trosterud, forthcoming) for a presentation of the North S\u00e1mi CG.10 https//svn.code.sf.net/p/apertium/svn/ nursery/apertium-sme-fin: revision 76019 11 https://victorio.uit.no/langtech/trunk/ langs/sme: revision 14746412 http://se.wikipedia.org", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| } |
| ], |
| "back_matter": [ |
| { |
| "text": "The work on this North S\u00e1mi-Finnish machine translation system was partially funded by the Google Summer of Code and Google Code-In programmes, and partly by a Norsk forsingsr\u00e5d grant (234299) on machine translation between S\u00e1mi languages.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Acknowledgements", |
| "sec_num": null |
| } |
| ], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "forthcoming. Ord sett innafra og utafra -en datalingvistisk analyse av nordsamisk", |
| "authors": [ |
| { |
| "first": "Lene", |
| "middle": [], |
| "last": "Antonsen", |
| "suffix": "" |
| }, |
| { |
| "first": "Trond", |
| "middle": [], |
| "last": "Trosterud", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Lene Antonsen and Trond Trosterud. forthcoming. Ord sett innafra og utafra -en datalingvistisk analyse av nordsamisk. Norsk Lingvistisk Tidsskrift.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "A North Saami to South Saami machine translation prototype", |
| "authors": [ |
| { |
| "first": "Lene", |
| "middle": [], |
| "last": "Antonsen", |
| "suffix": "" |
| }, |
| { |
| "first": "Trond", |
| "middle": [], |
| "last": "Trosterud", |
| "suffix": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [], |
| "last": "Tyers", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "", |
| "volume": "4", |
| "issue": "", |
| "pages": "11--27", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Lene Antonsen, Trond Trosterud, and Francis Tyers. 2016. A North Saami to South Saami machine trans- lation prototype. 4:11-27.", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "Re-evaluation the role of bleu in machine translation research", |
| "authors": [ |
| { |
| "first": "Chris", |
| "middle": [], |
| "last": "Callison", |
| "suffix": "" |
| }, |
| { |
| "first": "-", |
| "middle": [], |
| "last": "Burch", |
| "suffix": "" |
| }, |
| { |
| "first": "Miles", |
| "middle": [], |
| "last": "Osborne", |
| "suffix": "" |
| }, |
| { |
| "first": "Philipp", |
| "middle": [], |
| "last": "Koehn", |
| "suffix": "" |
| } |
| ], |
| "year": 2006, |
| "venue": "EACL", |
| "volume": "6", |
| "issue": "", |
| "pages": "249--256", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Chris Callison-Burch, Miles Osborne, and Philipp Koehn. 2006. Re-evaluation the role of bleu in machine translation research. In EACL, volume 6, pages 249-256.", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "Gema Ram\u00edrez-S\u00e1nchez, and Francis M Tyers", |
| "authors": [ |
| { |
| "first": "Mireia", |
| "middle": [], |
| "last": "Mikel L Forcada", |
| "suffix": "" |
| }, |
| { |
| "first": "Jacob", |
| "middle": [], |
| "last": "Ginest\u00ed-Rosell", |
| "suffix": "" |
| }, |
| { |
| "first": "Jim", |
| "middle": [ |
| "O" |
| ], |
| "last": "Nordfalk", |
| "suffix": "" |
| }, |
| { |
| "first": "Sergio", |
| "middle": [], |
| "last": "'regan", |
| "suffix": "" |
| }, |
| { |
| "first": "Juan", |
| "middle": [ |
| "Antonio" |
| ], |
| "last": "Ortiz-Rojas", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "P\u00e9rez-Ortiz", |
| "suffix": "" |
| } |
| ], |
| "year": 2011, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Mikel L Forcada, Mireia Ginest\u00ed-Rosell, Jacob Nord- falk, Jim O'Regan, Sergio Ortiz-Rojas, Juan An- tonio P\u00e9rez-Ortiz, Felipe S\u00e1nchez-Mart\u00ednez, Gema Ram\u00edrez-S\u00e1nchez, and Francis M Tyers. 2011.", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "Apertium: a free/open-source platform for rulebased machine translation. Machine translation", |
| "authors": [], |
| "year": null, |
| "venue": "", |
| "volume": "25", |
| "issue": "", |
| "pages": "127--144", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Apertium: a free/open-source platform for rule- based machine translation. Machine translation, 25(2):127-144.", |
| "links": null |
| }, |
| "BIBREF5": { |
| "ref_id": "b5", |
| "title": "Constraint Grammar: a language-independent system for parsing unrestricted text", |
| "authors": [ |
| { |
| "first": "Fred", |
| "middle": [], |
| "last": "Karlsson", |
| "suffix": "" |
| }, |
| { |
| "first": "Atro", |
| "middle": [], |
| "last": "Voutilainen", |
| "suffix": "" |
| }, |
| { |
| "first": "Juha", |
| "middle": [], |
| "last": "Heikkilae", |
| "suffix": "" |
| }, |
| { |
| "first": "Arto", |
| "middle": [], |
| "last": "Anttila", |
| "suffix": "" |
| } |
| ], |
| "year": 1995, |
| "venue": "", |
| "volume": "4", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Fred Karlsson, Atro Voutilainen, Juha Heikkilae, and Arto Anttila. 1995. Constraint Grammar: a language-independent system for parsing unre- stricted text, volume 4. Walter de Gruyter.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "Constraint grammar as a framework for parsing running text", |
| "authors": [ |
| { |
| "first": "Fred", |
| "middle": [], |
| "last": "Karlsson", |
| "suffix": "" |
| } |
| ], |
| "year": 1990, |
| "venue": "Proceedings of the 13th conference on Computational linguistics", |
| "volume": "3", |
| "issue": "", |
| "pages": "168--173", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Fred Karlsson. 1990. Constraint grammar as a frame- work for parsing running text. In Proceedings of the 13th conference on Computational linguistics- Volume 3, pages 168-173. Association for Compu- tational Linguistics.", |
| "links": null |
| }, |
| "BIBREF7": { |
| "ref_id": "b7", |
| "title": "Two-level morphology-A General Computational Model for Word-Form Recognition and Production", |
| "authors": [ |
| { |
| "first": "Kimmo", |
| "middle": [], |
| "last": "Koskenniemi", |
| "suffix": "" |
| } |
| ], |
| "year": 1983, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kimmo Koskenniemi. 1983. Two-level morphology- A General Computational Model for Word-Form Recognition and Production. Ph.D. thesis, Depart- ment of General Linguistics. University of Helsinki, Finland.", |
| "links": null |
| }, |
| "BIBREF8": { |
| "ref_id": "b8", |
| "title": "Binary codes capable of correcting deletions, insertions, and reversals", |
| "authors": [ |
| { |
| "first": "", |
| "middle": [], |
| "last": "Vladimir I Levenshtein", |
| "suffix": "" |
| } |
| ], |
| "year": 1966, |
| "venue": "Soviet physics doklady", |
| "volume": "10", |
| "issue": "", |
| "pages": "707--710", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Vladimir I Levenshtein. 1966. Binary codes capable of correcting deletions, insertions, and reversals. In Soviet physics doklady, volume 10, pages 707-710.", |
| "links": null |
| }, |
| "BIBREF9": { |
| "ref_id": "b9", |
| "title": "Hfst tools for morphology-an efficient open-source package for construction of morphological analyzers", |
| "authors": [ |
| { |
| "first": "Miikka", |
| "middle": [], |
| "last": "Krister Lind\u00e9n", |
| "suffix": "" |
| }, |
| { |
| "first": "Tommi", |
| "middle": [], |
| "last": "Silfverberg", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Pirinen", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "International Workshop on Systems and Frameworks for Computational Morphology", |
| "volume": "", |
| "issue": "", |
| "pages": "28--47", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Krister Lind\u00e9n, Miikka Silfverberg, and Tommi Piri- nen. 2009. Hfst tools for morphology-an efficient open-source package for construction of morpholog- ical analyzers. In International Workshop on Sys- tems and Frameworks for Computational Morphol- ogy, pages 28-47. Springer.", |
| "links": null |
| }, |
| "BIBREF10": { |
| "ref_id": "b10", |
| "title": "Building an open-source development infrastructure for language technology projects", |
| "authors": [ |
| { |
| "first": "Tommi", |
| "middle": [ |
| "A" |
| ], |
| "last": "Sjur N Moshagen", |
| "suffix": "" |
| }, |
| { |
| "first": "Trond", |
| "middle": [], |
| "last": "Pirinen", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Trosterud", |
| "suffix": "" |
| } |
| ], |
| "year": 2013, |
| "venue": "Proceedings of the 19th Nordic Conference of Computational Linguistics", |
| "volume": "16", |
| "issue": "", |
| "pages": "343--352", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Sjur N Moshagen, Tommi A Pirinen, and Trond Trosterud. 2013. Building an open-source de- velopment infrastructure for language technology projects. In Proceedings of the 19th Nordic Con- ference of Computational Linguistics (NODALIDA 2013); May 22-24; 2013; Oslo University; Norway. NEALT Proceedings Series 16, number 085, pages 343-352. Link\u00f6ping University Electronic Press.", |
| "links": null |
| }, |
| "BIBREF11": { |
| "ref_id": "b11", |
| "title": "Development and use of computational morphology of finnish in the open source and open science era: Notes on experiences with omorfi development", |
| "authors": [ |
| { |
| "first": "", |
| "middle": [], |
| "last": "Tommi A Pirinen", |
| "suffix": "" |
| } |
| ], |
| "year": 2015, |
| "venue": "SKY Journal of Linguistics", |
| "volume": "28", |
| "issue": "", |
| "pages": "381--393", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Tommi A Pirinen. 2015. Development and use of com- putational morphology of finnish in the open source and open science era: Notes on experiences with omorfi development. SKY Journal of Linguistics, 28:381-393.", |
| "links": null |
| }, |
| "BIBREF12": { |
| "ref_id": "b12", |
| "title": "A free/open-source kazakh-tatar machine translation system", |
| "authors": [ |
| { |
| "first": "Ilnar", |
| "middle": [], |
| "last": "Salimzyanov", |
| "suffix": "" |
| }, |
| { |
| "first": "F", |
| "middle": [], |
| "last": "Washington", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Tyers", |
| "suffix": "" |
| } |
| ], |
| "year": 2013, |
| "venue": "Machine Translation Summit XIV", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ilnar Salimzyanov, J Washington, and F Tyers. 2013. A free/open-source kazakh-tatar machine translation system. Machine Translation Summit XIV.", |
| "links": null |
| }, |
| "BIBREF13": { |
| "ref_id": "b13", |
| "title": "Bleu is not the colour: How optimising bleu reduces translation quality", |
| "authors": [ |
| { |
| "first": "Aaron", |
| "middle": [], |
| "last": "Smith", |
| "suffix": "" |
| }, |
| { |
| "first": "Christian", |
| "middle": [], |
| "last": "Hardmeier", |
| "suffix": "" |
| }, |
| { |
| "first": "J\u00f6rg", |
| "middle": [], |
| "last": "Tiedemann", |
| "suffix": "" |
| } |
| ], |
| "year": 2014, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Aaron Smith, Christian Hardmeier, and J\u00f6rg Tiede- mann. 2014. Bleu is not the colour: How optimising bleu reduces translation quality.", |
| "links": null |
| }, |
| "BIBREF14": { |
| "ref_id": "b14", |
| "title": "Climbing mount bleu: The strange world of reachable high-bleu translations", |
| "authors": [ |
| { |
| "first": "Aaron", |
| "middle": [], |
| "last": "Smith", |
| "suffix": "" |
| }, |
| { |
| "first": "Christian", |
| "middle": [], |
| "last": "Hardmeier", |
| "suffix": "" |
| }, |
| { |
| "first": "J\u00f6rg", |
| "middle": [], |
| "last": "Tiedemann", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "Baltic Journal of Modern Computing", |
| "volume": "4", |
| "issue": "2", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Aaron Smith, Christian Hardmeier, and J\u00f6rg Tiede- mann. 2016. Climbing mount bleu: The strange world of reachable high-bleu translations. Baltic Journal of Modern Computing, 4(2):269.", |
| "links": null |
| }, |
| "BIBREF15": { |
| "ref_id": "b15", |
| "title": "Evaluating North S\u00e1mi to Norwegian assimilation RBMT", |
| "authors": [ |
| { |
| "first": "Trond", |
| "middle": [], |
| "last": "Trosterud", |
| "suffix": "" |
| }, |
| { |
| "first": "Kevin", |
| "middle": [ |
| "Brubeck" |
| ], |
| "last": "Unhammer", |
| "suffix": "" |
| } |
| ], |
| "year": 2012, |
| "venue": "Proceedings of the Third International Workshop on Free/Open-Source Rule-Based Machine Translation", |
| "volume": "3", |
| "issue": "", |
| "pages": "13--26", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Trond Trosterud and Kevin Brubeck Unhammer. 2013. Evaluating North S\u00e1mi to Norwegian assimilation RBMT. In Proceedings of the Third International Workshop on Free/Open-Source Rule-Based Ma- chine Translation (FreeRBMT 2012), volume 3 of Technical report, pages 13-26. Department of Com- puter Science and Engineering, Chalmers University of Technology and University of Gothenburg.", |
| "links": null |
| }, |
| "BIBREF16": { |
| "ref_id": "b16", |
| "title": "Disambiguering av homonymi i Nord-og Lulesamisk. Suomalais-Ugrilaisen Seuran Toimituksia = M\u00e9moires de la Soci\u00e9t\u00e9 Fi nno-Ougrienne. S\u00e1mit, s\u00e1nit, s\u00e1tneh\u00e1mit. Riepmo\u010d\u00e1la Pekka Sammallahtii miessem\u00e1nu 21", |
| "authors": [ |
| { |
| "first": "Trond", |
| "middle": [], |
| "last": "Trosterud", |
| "suffix": "" |
| }, |
| { |
| "first": "Linda", |
| "middle": [], |
| "last": "Wiechetek", |
| "suffix": "" |
| } |
| ], |
| "year": 2007, |
| "venue": "", |
| "volume": "253", |
| "issue": "", |
| "pages": "375--395", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Trond Trosterud and Linda Wiechetek. 2007. Dis- ambiguering av homonymi i Nord-og Lulesamisk. Suomalais-Ugrilaisen Seuran Toimituksia = M\u00e9- moires de la Soci\u00e9t\u00e9 Fi nno-Ougrienne. S\u00e1mit, s\u00e1nit, s\u00e1tneh\u00e1mit. Riepmo\u010d\u00e1la Pekka Sammallahtii miessem\u00e1nu 21. beaivve 2007, 253:375-395.", |
| "links": null |
| }, |
| "BIBREF17": { |
| "ref_id": "b17", |
| "title": "Developing prototypes for machine translation between two S\u00e1mi languages", |
| "authors": [ |
| { |
| "first": "Francis", |
| "middle": [], |
| "last": "Tyers", |
| "suffix": "" |
| }, |
| { |
| "first": "Linda", |
| "middle": [], |
| "last": "Wiechetek", |
| "suffix": "" |
| }, |
| { |
| "first": "Trond", |
| "middle": [], |
| "last": "Trosterud", |
| "suffix": "" |
| } |
| ], |
| "year": 2009, |
| "venue": "Proceedings of the 13th Annual Conference of the European Association for Machine Translation", |
| "volume": "09", |
| "issue": "", |
| "pages": "120--128", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Francis Tyers, Linda Wiechetek, and Trond Trosterud. 2009. Developing prototypes for machine transla- tion between two S\u00e1mi languages. In Proceedings of the 13th Annual Conference of the European As- sociation for Machine Translation, EAMT09, pages 120-128.", |
| "links": null |
| }, |
| "BIBREF18": { |
| "ref_id": "b18", |
| "title": "Shooting at flies in the dark: Rule-based lexical selection for a minority language pair", |
| "authors": [ |
| { |
| "first": "Linda", |
| "middle": [], |
| "last": "Wiechetek", |
| "suffix": "" |
| }, |
| { |
| "first": "Francis", |
| "middle": [], |
| "last": "Tyers", |
| "suffix": "" |
| }, |
| { |
| "first": "Thomas", |
| "middle": [], |
| "last": "Omma", |
| "suffix": "" |
| } |
| ], |
| "year": 2010, |
| "venue": "Lecture Notes in Artificial Intelligence", |
| "volume": "6233", |
| "issue": "", |
| "pages": "418--429", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Linda Wiechetek, Francis Tyers, and Thomas Omma. 2010. Shooting at flies in the dark: Rule-based lex- ical selection for a minority language pair. Lecture Notes in Artificial Intelligence, 6233:418-429.", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "FIGREF0": { |
| "uris": null, |
| "num": null, |
| "text": "Apertium structure (Image from apertium wiki by user Rcrowther) http://wiki.apertium. org/wiki/Workflow_diagram North S\u00e1mi input: S\u00e1megielat leat gielat maid s\u00e1mit h\u00e1llet.", |
| "type_str": "figure" |
| }, |
| "FIGREF1": { |
| "uris": null, |
| "num": null, |
| "text": "Example entries from the bilingual transfer lexicon. Finnish is on the right, and North S\u00e1mi on the left.", |
| "type_str": "figure" |
| }, |
| "TABREF0": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "text": "Translation process for the North S\u00e1mi phrase S\u00e1megielat leat gielat maid s\u00e1mit h\u00e1llet (The S\u00e1mi languages are the languages that the S\u00e1mis speak)", |
| "content": "<table/>" |
| }, |
| "TABREF2": { |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "text": "Word error rate over the corpora; OOV is the number of out-of-vocabulary (unknown) words.", |
| "content": "<table/>" |
| } |
| } |
| } |
| } |
