Grouping Synonyms by Definitions

29547

Authors

nan

Abstract

We present a method for grouping the synonyms of a lemma according to its dictionary senses. The senses are defined by a large machine read- able dictionary for French, the TLFi (Tr´esor de la langue fran¸caise informatis´e) and the syn- onyms are given by 5 synonym dictionaries (also for French). To evaluate the proposed method, we manually constructed a gold standard where for each (word, definition) pair and given the set of synonyms defined for that word by the 5 synonym dictionaries, 4 lexicographers speci- fied the set of synonyms they judge adequate. While inter-annotator agreement ranges on that task from 67% to at best 88% depending on the annotator pair and on the synonym dictionary being considered, the automatic procedure we propose scores a precision of 67% and a recall of 71%. The proposed method is compared with related work namely, word sense disambiguation, synonym lexicon acquisition and WordNet con- struction. Keywords Similarity measures, Synonyms, Lexical Acquisition 1

Keywords

nan

Full Content

International Conference RANLP 2009 - Borovets, Bulgaria, pages 76–81 Grouping synonyms by definitions Ingrid Falk∗ INRIA / Universit´e Nancy 2 ingrid.falk@loria.fr Claire Gardent CNRS / LORIA, Nancy claire.gardent@loria.fr Evelyne Jacquey CNRS / ATILF, Nancy evelyne.jacquey@atilf.fr Fabienne Venant Universit´e Nancy 2 / LORIA, Nancy fabienne.venant@loria.fr Abstract We present a method for grouping the synonyms of a lemma according to its dictionary senses. The senses are defined by a large machine read- able dictionary for French, the TLFi (Tr´esor de la langue fran¸caise informatis´e) and the syn- onyms are given by 5 synonym dictionaries (also for French). To evaluate the proposed method, we manually constructed a gold standard where for each (word, definition) pair and given the set of synonyms defined for that word by the 5 synonym dictionaries, 4 lexicographers speci- fied the set of synonyms they judge adequate. While inter-annotator agreement ranges on that task from 67% to at best 88% depending on the annotator pair and on the synonym dictionary being considered, the automatic procedure we propose scores a precision of 67% and a recall of 71%. The proposed method is compared with related work namely, word sense disambiguation, synonym lexicon acquisition and WordNet con- struction. Keywords Similarity measures, Synonyms, Lexical Acquisition 1 Introduction Synonymic resources for French are still limited in scope, quality and/or availability. Thus the French WordNet (Frewn) created within the EuroWordNet project [16] has limited scope (3 777 verbs and 14 618 nouns vs. 7 384 verbs and 42 849 nouns in the morpho- logical lexicon for French Morphalou) and has not been widely used mainly due to licensing issues. The alter- native open-source WordNet for French called Wolf (WordNet Libre du Francais, [24]) remedies the first shortcoming (restrictive licensing) and aims to achieve a wider coverage by automating the WordNet con- struction process using an extend approach which in essence, translates the synsets from Princeton Word- Net (PWN) into French. However, compared to Mor- phalou, Wolf is still incomplete (979 verbs and 34 827 nouns). Finally, the synonym lexicon DicoSyn [19] ∗The research presented in this paper was partially supported by the TALC theme of the CPER ”Mod´elisation, Information et Syst`emes Num´eriques” funded by the R´egion Lorraine. We also gratefully acknowledge the ATILF for letting us access their synonym database and the TLFi. is restricted to assigning sets of synonyms to lemmas thereby lacking both categorial information and defi- nitions. In this paper, we present a method for grouping syn- onyms by senses and evaluate it on the synonyms given by 5 synonym dictionaries included in the Atilf syn- onym database. The long term aim is to apply this method to these synonym dictionaries so as to build a uniform synonymic resource for French in which each lemma is assigned a part of speech, a set of (TLFi) definitions and for each given definition, a set of syn- onyms. The resulting resource should complement Di- coSyn and Wolf. Contrary to DicoSyn, it will in- clude categorial information and associate groups of synonyms with definitions. …

Content truncated to 500 words

Labeled Text

We present a method for grouping the synonyms of a lemma according to its dictionary senses. The senses are defined by a large machine readable dictionary for French, the TLFi (Tr´esor de la langue fran¸caise informatis´e) and the synonyms are given by 5 synonym dictionaries (also for French). To evaluate the proposed method, we manually constructed a gold standard where for each (word, definition) pair and given the set of synonyms defined for that word by the 5 synonym dictionaries, 4 lexicographers specified the set of synonyms they judge adequate. While inter-annotator agreement ranges on that task from 67% to at best 88% depending on the annotator pair and on the synonym dictionary being considered, the automatic procedure we propose scores a precision of 67% and a recall of 71%. The proposed method is compared with related work namely, word sense disambiguation, synonym lexicon acquisition and WordNet construction. Keywords Similarity measures, Synonyms, Lexical Acquisition 1

Publication Details

Paper ID:
29547
Year:
2009
Publication:
TIPSTER
Publication Type:
conf
DOI:
nan