WordNet is a lexical database of English where words are grouped into sets of synonyms (synsets), each expressing a distinct concept. Synsets are interlinked by means of conceptual-semantic and lexical relations. WordNet has turned out to be an indispensable resource in the processing of natural language, and based on its model similar lexical databases were created for many other languages.
However, constructing such databases takes many years of work and is very costly. On the other hand, methods for the automatic identification of semantically related words based on large text corpora have reached a considerable degree of maturity, with the results coming close to native speakers’ performance. The proposed project aims at further refining and extending these approaches, thereby making it possible to fully automatically generate a resource similar to WordNet. The developed system will be largely language independent and is to be applied to four European languages, namely English, French, German, and Spanish. The resulting databases will be made freely available on the internet.
This is an outline of the proposed methodology: Starting from a part-of-speech tagged corpus, various methods for computing related words, such as syntax-based or utilizing latent semantic analysis, are applied and the results are systematically compared. The quality is evaluated by comparing the simulation results to a recently published data set comprising the 200,000 human similarity judgments from the Princeton Evocation project, rather than to the well established but inadequate 80 item TOEFL dataset. To identify synsets, an algorithm for unsupervised word sense induction is applied, and each word in the vocabulary is assigned to one or (if ambiguous) several of the synsets. Finally, to determine the relations between words (e.g. synonymy, hyponymy, holonymy, meronymy), an adapted version of Peter Turney’s approach for computing relational similarities is developed and applied.
Call for proposal
See other projects for this call