| { |
| "paper_id": "Y98-1035", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T13:37:32.944402Z" |
| }, |
| "title": "Using A Semantic Classification in Parsing Chinese: Some Preliminary Results", |
| "authors": [ |
| { |
| "first": "Wee", |
| "middle": [], |
| "last": "Kok", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Hong Kong University of Science & Technology Clear", |
| "location": { |
| "addrLine": "Water Bay Road", |
| "settlement": "Kowloon, Hong Kong" |
| } |
| }, |
| "email": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Gan", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Hong Kong University of Science & Technology Clear", |
| "location": { |
| "addrLine": "Water Bay Road", |
| "settlement": "Kowloon, Hong Kong" |
| } |
| }, |
| "email": "" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "This paper describes a semantic classification of Chinese and compares its performance with CKIP's word-class classification on the task of identifying the attributive dependency relation and the unmarked coordinative dependency relation of word bigrams made up from nouns, verbs, and adjectives. \u2022 ActantlRel The actantl relation, denoted as ActantlRel, connects between a lexeme whose meaning is a functor (the head) and its first argument (the dependent). Examples include:1 ActantlRel(gg 'agree', ft 'you') ActantlRel(t 'big', AR 'tolerance') \u2022 ActantlRel(&)4J 'useless', g 'say') Actant2Rel ActantlRel(ki 'is', lb* `deligene) The actant2 relation, denoted as Actant2Rel, connects between a lexeme whose meaning is a functor (the head) and its second argument (the dependent). Examples include: Actant2Rel('drink',-,1 \u00f7; 'tea') Actant2Rel(SY1 'feel', aVj 'cordial') Actant2Rel(ge 'like', Mil< 'swim') \u2022 MotionStateRel The motion-state relation, denoted as MotionStateRel, exists between two predicates, the dependent (a predicate) expresses the result of the motion conveyed by the other predicate (the head). MotionStateRel(* 'take', A 'walk') MotionStateRel(IR 'make', t 'big') MotionStateRel(4 'sit', *E 'upright') \u2022 AttrRel The attributive relation, denoted as AttrRel, covers all kinds of modifiers: AttrRel(fFT 'cup', 3g34 'glass') AttrRel(44 'dictionary', Ift 'pocket-size') AttrRel(,' a 'construct', I 'industry') AttrRel(EN 'regular', 'not') AttrRel('M 'method', Via 'train') AttrRel(gM 'extend', 4V 'stay') \u2022 CoordRel The coordinative relation discussed here, denoted as CoordRel, focuses only on unmarked coordinate constructions without any conjunction. The left-hand conjunct is assumed as the head. Examples are: CoordRel(*A 'wood table, 710* 'wood chair') CoordRe1CFW 'rain', ttUR 'wind blow') CoordRel(1)] 'lively', r&rit 'vigorously') In this paper, we propose a semantic classification of Chinese. We compare its performance against the", |
| "pdf_parse": { |
| "paper_id": "Y98-1035", |
| "_pdf_hash": "", |
| "abstract": [ |
| { |
| "text": "This paper describes a semantic classification of Chinese and compares its performance with CKIP's word-class classification on the task of identifying the attributive dependency relation and the unmarked coordinative dependency relation of word bigrams made up from nouns, verbs, and adjectives. \u2022 ActantlRel The actantl relation, denoted as ActantlRel, connects between a lexeme whose meaning is a functor (the head) and its first argument (the dependent). Examples include:1 ActantlRel(gg 'agree', ft 'you') ActantlRel(t 'big', AR 'tolerance') \u2022 ActantlRel(&)4J 'useless', g 'say') Actant2Rel ActantlRel(ki 'is', lb* `deligene) The actant2 relation, denoted as Actant2Rel, connects between a lexeme whose meaning is a functor (the head) and its second argument (the dependent). Examples include: Actant2Rel('drink',-,1 \u00f7; 'tea') Actant2Rel(SY1 'feel', aVj 'cordial') Actant2Rel(ge 'like', Mil< 'swim') \u2022 MotionStateRel The motion-state relation, denoted as MotionStateRel, exists between two predicates, the dependent (a predicate) expresses the result of the motion conveyed by the other predicate (the head). MotionStateRel(* 'take', A 'walk') MotionStateRel(IR 'make', t 'big') MotionStateRel(4 'sit', *E 'upright') \u2022 AttrRel The attributive relation, denoted as AttrRel, covers all kinds of modifiers: AttrRel(fFT 'cup', 3g34 'glass') AttrRel(44 'dictionary', Ift 'pocket-size') AttrRel(,' a 'construct', I 'industry') AttrRel(EN 'regular', 'not') AttrRel('M 'method', Via 'train') AttrRel(gM 'extend', 4V 'stay') \u2022 CoordRel The coordinative relation discussed here, denoted as CoordRel, focuses only on unmarked coordinate constructions without any conjunction. The left-hand conjunct is assumed as the head. Examples are: CoordRel(*A 'wood table, 710* 'wood chair') CoordRe1CFW 'rain', ttUR 'wind blow') CoordRel(1)] 'lively', r&rit 'vigorously') In this paper, we propose a semantic classification of Chinese. We compare its performance against the", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Abstract", |
| "sec_num": null |
| } |
| ], |
| "body_text": [ |
| { |
| "text": "In parsing Chinese, the practice of deriving the semantic structure of a sentence from word-class information faces two main challenges: (i) it is difficult to determine word classes due to the lack of inflections in Chinese; (ii) there is a lot of ambiguities in the mapping of word classes to the grammatical roles and to the semantic structure. Table 1 summarizes the types of dependency relation that could exist in word bigrams composed from nouns, verbs, and adjectives. traditional approach on the task of identifying the attributive and the unmarked coordinative relations in word bigrams made up from nouns, verbs, and adjectives.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 348, |
| "end": 355, |
| "text": "Table 1", |
| "ref_id": "TABREF1" |
| } |
| ], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "We propose a semantic classification of Chinese. This classification is derived by adopting the taxonomy of linguistic meanings proposed in the Meaning-Text approach (Polguere, to appear) and developing it further based on (Dong, 1994; 14:13M-Mra, 1993) . Figures 1, 2, 3, 4, and 5 shows the details.", |
| "cite_spans": [ |
| { |
| "start": 166, |
| "end": 187, |
| "text": "(Polguere, to appear)", |
| "ref_id": null |
| }, |
| { |
| "start": 223, |
| "end": 235, |
| "text": "(Dong, 1994;", |
| "ref_id": "BIBREF0" |
| }, |
| { |
| "start": 236, |
| "end": 253, |
| "text": "14:13M-Mra, 1993)", |
| "ref_id": null |
| }, |
| { |
| "start": 256, |
| "end": 266, |
| "text": "Figures 1,", |
| "ref_id": null |
| }, |
| { |
| "start": 267, |
| "end": 269, |
| "text": "2,", |
| "ref_id": null |
| }, |
| { |
| "start": 270, |
| "end": 272, |
| "text": "3,", |
| "ref_id": null |
| }, |
| { |
| "start": 273, |
| "end": 275, |
| "text": "4,", |
| "ref_id": null |
| }, |
| { |
| "start": 276, |
| "end": 281, |
| "text": "and 5", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The Semantic Classification of Chinese", |
| "sec_num": "2" |
| }, |
| { |
| "text": "The object name and predicate classifications are adapted from EMICS (Dong, 1994) . The difference is that our classifications are not as fine-grained as EMICS. We prefer to verify the adequacy of our classification with actual applications before working on a finer classification. The quantifier and operator classifications are adapted from 1993). The most general concept of quantifier is used here. Words of measure, for example, are interpreted as quantifiers. Their first actant being a lexical meaning. It is the quantified element as well as the dominant node of the utterance meaning. The second actant is an utterance meaning containing the first actant. Figure 6 shows an example of the graphical representation of a quantifier. Operators are words having utterances as the actancts. Figure 7 shows an example.", |
| "cite_spans": [ |
| { |
| "start": 69, |
| "end": 81, |
| "text": "(Dong, 1994)", |
| "ref_id": "BIBREF0" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 666, |
| "end": 674, |
| "text": "Figure 6", |
| "ref_id": null |
| }, |
| { |
| "start": 796, |
| "end": 804, |
| "text": "Figure 7", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "The Semantic Classification of Chinese", |
| "sec_num": "2" |
| }, |
| { |
| "text": "Eight articles on the theme of travelling from the Sinica Corpus V1.0 ( r c-Mr (ga, 1995) are used in the study. They comprises a total of 5882 Chinese characters. There are 46 classes in the CKIP tagset. This tagset is derived from (11:13HZIa*, 1993) . In this paper, it is sufficient to simply state that tags which begin with \"N\" are noun classes; tags which begin with \"V\" are verb classes; non:predicative adjective is \"A\". Details about the CKIP tagset can be found in (E173a9Z1a*, 1995) . Word bigrams with word-class equals to noun, verb, and adjective are extracted. These bigrams are manually examined. Those with either an attributive relation or coordinative relation are retained.", |
| "cite_spans": [ |
| { |
| "start": 79, |
| "end": 89, |
| "text": "(ga, 1995)", |
| "ref_id": null |
| }, |
| { |
| "start": 233, |
| "end": 251, |
| "text": "(11:13HZIa*, 1993)", |
| "ref_id": null |
| }, |
| { |
| "start": 475, |
| "end": 493, |
| "text": "(E173a9Z1a*, 1995)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Word-Class Distributions", |
| "sec_num": "3" |
| }, |
| { |
| "text": "In Chinese, the vast majority of adjectives may function as the head of a verb phrase (Li and Thompson, 1981) . This type of adjectives has been classified as verbs in CKIP. Adjectives which cannot be the head of a verb phrase are called non-predicative adjective (4MMI7) in CKIP, and they are denoted by the tag \"A\". Examples are: 'big-scale', Mt 'positive', * 'newly properous', NE 'truly', AO, 'permanent', 'most properous', and _EAT 'best'. For verbs which could be used in a position that expects a nominal, they are not regarded as having two word classes: verb and noun. CKIP treats them as verbs. When they appear in a nominal position, they are tagged with an additional grammatical feature \"+nom\". For example, in the following bigrams:", |
| "cite_spans": [ |
| { |
| "start": 86, |
| "end": 109, |
| "text": "(Li and Thompson, 1981)", |
| "ref_id": "BIBREF1" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Word-Class Distributions", |
| "sec_num": "3" |
| }, |
| { |
| "text": "(1) ONI/VC+nom PEjr,/ Na (\"visit ticket\") (2) le / VC+nom /Na (\"manage method\") In (1) and (2), the verbs OE 'visit' and 'manage' are used as modifiers to a noun. They are therefore tagged with an addition grammatical feature `+nom'. Though CKIP has classified adjectives which may function as the head of a verb phrase as verbs, this class of verbs is handled differently with respect to the issue of nominalization. For example, in the following bigrams:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Word-Class Distributions", |
| "sec_num": "3" |
| }, |
| { |
| "text": "(3) RAN / VH A\u00b1/Na (\"handicap people\") (4) aff /VH f /Na (\"friendly country\") the verbs RN 'handicap' and Aff 'friendly' are not tagged with the feature `+nom' . These verbs would traditionally be regarded as adjectives. From the different treatment of verbs of this class (VH) from verbs in the previous class (VC) when both function as modifiers to nouns, CKIP does make an implicit distinction between them.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Word-Class Distributions", |
| "sec_num": "3" |
| }, |
| { |
| "text": "Based on the word-class distributions of bigrams with attributive and coordinative relations, the following observations are noted.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Word-Class Distributions", |
| "sec_num": "3" |
| }, |
| { |
| "text": "An attributive relation or a coordinative relation could exist in a word bigram with word class equals t\u00f2 VH VA'.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022", |
| "sec_num": null |
| }, |
| { |
| "text": "CoordReM5NH 'clever', a NA 'dedicate') AttrRel(AMNA 'tour', M . A/VH 'deep into')", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022", |
| "sec_num": null |
| }, |
| { |
| "text": "\u2022 It is not clear why verbs such as frff 'camp' and eg 'rest' in the following bigrams are tagged with the grammatical feature `-i-nom' while verbs such as VT7 'tour' and my6 'sightseeing' are not, even though all of them have the same word class 'VA' and all of them appear in bigrams with an attributive relation. Similar cases are also observed for verbs in the VJ and VK class. AttrRel(/Na 'certificate', afRvA 'tour') AttrRel(gn /Na 'purchase team', /VA 'sightseeing') AttrRel(*/Na 'car', Xt NA+nom 'camp') AttrRel(XW/Na 'need', SANA+nom 'rest')", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022", |
| "sec_num": null |
| }, |
| { |
| "text": "\u2022 A difficult issue in automatic word class tagging lies in nominalization, as exemplifed in the following examples.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022", |
| "sec_num": null |
| }, |
| { |
| "text": "AttrRel(F9X/Na 'ticket', OSINA+nom 'visit') Actant2Rel(f, tENA 'visit', Il1J/Nc 'musuem') AttrRel(*a/Na 'car type', NA+nom 'drive') Actant2Rel(M Na 'drive', ;.:M*/Na 'van') It is not clear how statistical information such as word ,cooccurrence frequency and word-class cooccurrence frequency are sufficient to decide whether nominalization exists in cases like P\u00d2 sightseeing' and WI'drive', since there is no inflection in Chinese. It is not clear how unmarked coordinative relation could be identified with word-class information alone. We observed that a coordinative relation could exist in a word bigram with two different subcategories of verb, such as VA VCL, VA VH, VC VA, VH VA, and VL VC. However, it is also possible for an attributive relation to exist in a bigram with two different subcategories of verb, such as VA VF, VA VB, VH VA, VH VC, VH VCL, and VH VK. A parser to identify both the attributive and coordinative relations in our list of word bigrams based on word-class information alone will have an approximate baseline recognition rate of 45.6% (41/90) and 21.2% (3/14) respectively: These figures are derived by considering only those cases without ambiguity. For example, all bigrams made up of (A Na), (A Nc) and (VH *) are attributive relations. All bigrams with its constituents having the same word-class are coordinative relations.", |
| "cite_spans": [ |
| { |
| "start": 35, |
| "end": 43, |
| "text": "'visit')", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022", |
| "sec_num": null |
| }, |
| { |
| "text": "The same list of word bigrams described in Section 3 are manually tagged with semantic classes. The following observations are noted.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semantic-Class Distributions", |
| "sec_num": "4." |
| }, |
| { |
| "text": "\u2022 The constituents of word bigrams with a coordinative relation have the same semantic class.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Semantic-Class Distributions", |
| "sec_num": "4." |
| }, |
| { |
| "text": "All attributive relations can be categorized as belonging to one of the following cases: a.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022", |
| "sec_num": null |
| }, |
| { |
| "text": "Property Value-X: The first constituent belongs to the semantic class jffi'til \"property value\", which describes some property value of the second constituent. The second constituent is labelled as \"X\", which means that there are no restrictions on the semantic class of the second constituent. There are 33 such cases. Examples include:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022", |
| "sec_num": null |
| }, |
| { |
| "text": "fjEft-t 'elegant-lady' ' 'big-have' s Aj5-g 'direct-process' ' 4=fi \u00a7 \u00a7- newly married-couple' ' 'diligently-guide' '", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022", |
| "sec_num": null |
| }, |
| { |
| "text": "'good-news' ' fiESffitil 'preferential-discount' )11P1J -JIM 'smoothly-obtain' ' A-g. 'doublebedroom' '-1111 'short-time' ' IR-ftfo l 'main-journey'.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022", |
| "sec_num": null |
| }, |
| { |
| "text": "X-Property: The second constituent describes some property of the first constituent. There are no restrictions on the semantic class of the first constituent. There are 8 such cases. i*,01-i*g 'rest-camp site' ' 'retired-official' ' 'remember old time-color' ' 'agree-degree'. d.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "Relation-X: The first constituent belongs to the semantic class,-, \"Relation\" and there are no restrictions on the semantic class of the second constituent. There are 6 such cases. Examples include:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "TR-1 'different-price' ' 41M-4 'relevant-document' Wi--4514* 'station out- organization' . e.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "Change of Relation-X or Change of State-X: The first constituent either belongs to the semantic class NM* \"Change of Relation\" or V#1 \u00a7 \"Change of State\". There are no restrictions on the semantic class of the second constituent. There are 11 such cases. Examples include:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "Act-Ii=in 'cross border-apply' ' 5:3M-EN '. washing-service'", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "'enquirytelephone' , 'shopping-discount'. f.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "The last category has the modifiers specifying the scope or range of the head. The most generic sense of scope and range is used here. There are 17 cases here.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "Ec*-8fijE. 'camping-activity area' ' $1*--U4V--1 'visit-purchase team' ' 11*-1111M", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "\u2022 \u2022 camping-tent' '", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "'apply-incident' ' CT --I gg. ff 'drive-route' ' tcg-S19 manage-department' ' -M 'manage-method' N,\" 'passport-signature' ' IN-N 'city-tour' ' -A)14 'international-valid' ' *Sq--ttifg, 'information-supply' guest room-service'. When all the word bigrams with attributive and coordinative relations are considered together with all the other bigrams with the ActantiRel and Actant2Rel relations (total of 256 bigrams), only two cases of ambuities are found.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "Actant2Rel(? / 'night stay/living', ca rail 'campsite/space') Actant2Re1( / 1 ,apply/do', *-M /tt `visa/reading') AttrRel( Z / 1115 'campsite/space', 171Kg, / 'rest/living') AttrRel(3C4/ Mt 'document/reading', EP/`apply/do') \u2022 Conceptually, it is possible that an ActantlRel exists in a bigram of the class `X-Property'. Let us assume that a parser is not able to identify all AttrRel relations in this class because of such ambiguity. Assuming further that the parser is also unable to identify AttrRel relations where the modifiers are generic scope/range of the head. The parser would still be able to achieve a recognition rate of 72.2% (65/90). Assuming that the simplistic strategy of assigning a coordinative relation if the constitutents of a bigram have the same semantic class. The accuracy of identifying coordinative relations in our sample will be 100%.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "b.", |
| "sec_num": null |
| }, |
| { |
| "text": "In this paper, we reported some preliminary results of using semantic classes to derive two dependency relations in Chinese word bigrams: the attributive and coordinative relations. We compared the results with the word classes used in the Sinica corpus. The results have suggested that a drastic improvement in parsing accuracy could be achieved with the semantic classification. Our second step now is to develop a statistical parsing model for Chinese based on this semantic classification. ", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Conclusion", |
| "sec_num": null |
| }, |
| { |
| "text": "In this paper, a dependency relation between two words is represented as: relation(head, dependent).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| } |
| ], |
| "back_matter": [ |
| { |
| "text": "This work has benefited from many discussions with Prof. Dong Zhendong. We would like to express our thanks to him.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Acknowledgements", |
| "sec_num": null |
| } |
| ], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "A Categorization System Of Chinese Movement Concepts", |
| "authors": [ |
| { |
| "first": "Zhendong", |
| "middle": [], |
| "last": "Dong", |
| "suffix": "" |
| } |
| ], |
| "year": 1994, |
| "venue": "International Conference on Chinese Computing '94", |
| "volume": "", |
| "issue": "", |
| "pages": "145--149", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Dong, Zhendong (1994) A Categorization System Of Chinese Movement Concepts. International Conference on Chinese Computing '94; page 145-149.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "Mandarin Chinese: a functional reference grammar", |
| "authors": [ |
| { |
| "first": "Charles", |
| "middle": [ |
| "N" |
| ], |
| "last": "Li", |
| "suffix": "" |
| }, |
| { |
| "first": "Sandra", |
| "middle": [ |
| "A" |
| ], |
| "last": "Thompson", |
| "suffix": "" |
| } |
| ], |
| "year": 1981, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Li, Charles N., Thompson, Sandra A. (1981) Mandarin Chinese: a functional reference grammar. University of California Press, 1981.", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "Meaning-Text Semantic Networks As A Formal Language", |
| "authors": [ |
| { |
| "first": "Alain", |
| "middle": [], |
| "last": "Polguere", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "Current Issues in Meaning-Text Linguistics", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Polguere, Alain (to appear) Meaning-Text Semantic Networks As A Formal Language. To appear in Current Issues in Meaning-Text Linguistics;", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "ci : TAi3cF5t1R*ItaieMniN4AM BA EI:IAF5tIR*PA*lif5ffifii.kMV*", |
| "authors": [ |
| { |
| "first": "", |
| "middle": [], |
| "last": "Emzum*", |
| "suffix": "" |
| } |
| ], |
| "year": 1995, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "95--97", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "EMZUM* (1995) ci : TAi3cF5t1R*ItaieMniN4AM BA EI:IAF5tIR*PA*lif5ffifii.kMV* 95- 02", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "FIGREF0": { |
| "num": null, |
| "type_str": "figure", |
| "uris": null, |
| "text": "Figure 3: Predicate Classification" |
| }, |
| "TABREF1": { |
| "html": null, |
| "text": "Examples o dependency relations", |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td>Dependency</td><td/><td>Word-Class</td><td/><td>Examples</td></tr><tr><td>Relation</td><td/><td/><td/></tr><tr><td>ActantlRel</td><td/><td>Noun Verb Verb Verb Noun Adjective Adjective Verb</td><td/><td>fircli-at 'you agree' Eivtg 'say useless' *At 'tolerance big' VAC% 'diligent is'</td></tr><tr><td>. Actant2Rel</td><td/><td>Verb Noun</td><td/><td>fir, 'drink tea'</td></tr><tr><td/><td>.</td><td>Verb Verb Verb Adjective</td><td/><td>gag* 'like swim' SMIVj 'feel cordial'</td></tr><tr><td>MotionStateRel</td><td/><td>Verb Verb</td><td/><td>*</td><td>'take walk'</td></tr><tr><td/><td/><td>Verb Adjective</td><td colspan=\"2\">rat 'make big' % Pt \u00a7.E 'sit upright'</td></tr><tr><td>AttrRel</td><td/><td>Noun Noun</td><td/><td>::-3ififf-T 'glass cup'</td></tr><tr><td/><td/><td>Noun Verb Verb Noun</td><td/><td>i*m=a 'industry construct' ViliMit 'train method'</td></tr><tr><td/><td/><td>Verb Verb</td><td/><td>.ggitg 'stay extend'</td></tr><tr><td/><td/><td>Adjective Noun</td><td colspan=\"2\">'::::: 'pocket-size dictionary'</td></tr><tr><td/><td colspan=\"2\">Adjective Adjective Adjective Verb</td><td/><td>4PER 'not regular' NA*g 'serious learn'</td></tr><tr><td>CoordRel</td><td/><td>Noun Noun</td><td colspan=\"2\">*A** 'wood table wood chair'</td></tr><tr><td>(unmarked)</td><td/><td>Verb Verb</td><td/><td>-F-Mtilat 'rain wind-blow'</td></tr><tr><td/><td colspan=\"2\">Adjective Adjective</td><td/><td>.NiVil 'lively vigorously'</td></tr><tr><td/><td colspan=\"2\">Lexical Meaning</td><td/><td>Utterance Meaning</td></tr><tr><td>Object Name</td><td/><td>Functor</td><td/></tr><tr><td colspan=\"2\">linking one or several lexical meanings</td><td/><td colspan=\"2\">linking several utterance meanings</td></tr><tr><td>Predicate</td><td colspan=\"3\">linking a lexical meaning and an utterance meaning containing this</td><td>Operator</td></tr><tr><td/><td colspan=\"2\">lexical meaning</td><td/></tr><tr><td/><td/><td>Quantifier</td><td/></tr><tr><td colspan=\"5\">Figure 1: Taxonomy of linguistic meanings</td></tr></table>" |
| } |
| } |
| } |
| } |
