About Inferences in a Crowdsourced Lexical-Semantic Network

Proceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics, pages 174–182, Gothenburg, Sweden, April 26-30 2014. c⃝2014 Association for Computational Linguistics About Inferences in a Crowdsourced Lexical-Semantic Network Manel Zarrouk UM2-LIRMM 161 rue Ada 34095 Montpellier, FRANCE manel.zarrouk@lirmm.fr Mathieu Lafourcade UM2-LIRMM 161 rue Ada 34095 Montpellier, FRANCE mathieu.lafourcade@lirmm.fr Alain Joubert UM2-LIRMM 161 rue Ada 34095 Montpellier, FRANCE alain.joubert@lirmm.fr Abstract Automatically inferring new relations from already existing ones is a way to improve the quality of a lexical network by relation densification and error de- tection. In this paper, we devise such an approach for the JeuxDeMots lexi- cal network, which is a freely avalaible lexical network for French. We first present deduction (generic to specific) and induction (specific to generic) which are two inference schemes ontologically founded. We then propose abduction as a third form of inference scheme, which exploits examples similar to a tar- get term. 1 Introduction Building resources for Computational Linguis- tics (CL) is of crucial interest. Most of exist- ing lexical-semantic networks have been built by hand (like for instance WordNet (Miller et al., 1990)) and, despite that tools are generally designed for consistency checking, the task re- mains time consuming and costly. Fully auto- mated approaches are generally limited to term co-occurrences as extracting precise semantic relations between terms from corpora remains really difficult. Meanwhile, crowdsourcing ap- proaches are flowering in CL especially with the advent of Amazon Mechanical Turk or in a broader scope Wikipedia and Wiktionary, to cite the most well-known examples. WordNet is such a lexical network, constructed by hand at great cost, based on synsets which can be roughly considered as concepts (Fellbaum, 1988). Eu- roWordnet (Vossen., 1998) a multilingual ver- sion of WordNet and WOLF (Sagot., 2008) a French version of WordNet, were built by auto- mated crossing of WordNet and other lexical re- sources along with some manual checking. Nav- igli (2010) constructed automatically BabelNet a large multilingual lexical network from term co- occurrences in Wikipedia. A lexical-semantic network can contain lem- mas, word forms and multi-word expressions as entry points (nodes) along with word meanings and concepts. The idea itself of word senses in the lexicographic tradition may be debatable in the context of resources for semantic analysis, and we generally prefer to consider word us- ages. A given polysemous word, as identified by locutors, has several usages that might dif- fer substantially from word senses as classically defined. A given usage can also in turn have several deeper refinements and the whole set of usages can take the form of a decision tree. For example, frigate can be a bird or a ship. A frigate>boat can be distinguished as a modern ship with missiles and radar or an ancient vessel with sails. In the context of a collaborative con- struction, such a lexical resource should be con- sidered as being constantly evolving and a gen- eral rule of thumb is to have no definite certi- tude about the state of an …

Content truncated to 500 words