Tailored Feature Extraction for Lexical Disambiguation of English Verbs Based on Corpus Pattern Analysis

We give a report on a detailed study of automatic lexical disambiguation of 30 sample English verbs. We were drawing on a lexicon of English verb patterns based on the Corpus Pattern Analysis (CPA), which is a novel lexicographic method that seeks to cluster verb uses according to the morpho-syntactic, lexical and semantic/pragmatic similarity of their contexts rather than to associate them with abstract semantic definitions. We have trained several statistical classifiers to recognize these patterns, using morpho-syntactic as well as semantic features. In this paper we mainly concentrate on the procedures for feature extraction and feature selection and their evaluation. We show that tailoring the features to the verbs respectively, as they are implicitly contained in the pattern definitions (explicitly described in the lexicon), has the potential to significantly improve the accuracy of supervised statistical classifiers. TITLE AND ABSTRACT IN CZECH Rysy šité na míru anglickým slovesům pro automatickou lexikální disambiguaci pomocí Corpus Pattern Analysis Předkládáme detailní studii automatické lexikální disambiguace na pilotním vzork