Mining Parsing Results for Lexical Correction: Toward a Complete Correction Process of Wide-Coverage Lexicons

The coverage of a parser depends mostly on the quality of the underlying grammar and lexicon. The development of a lexicon both complete and accurate is an intricate and demanding task. We introduce a automatic process for detecting missing, incomplete and erroneous entries in a morphological and syntactic lexicon, and for suggesting corrections hypotheses for these entries. The detection of dubious lexical entries is tackled by two different techniques; the first one is based on a specific statistical model, the other one benefits from information provided by a part-of-speech tagger. The generation of correction hypotheses for dubious lexical entries is achieved by studying which modifications could improve the successful parse rate of sentences in which they occur. This process brings together various techniques based on taggers, parsers and statistical models. We report on its application for improving a large-coverage morphological and syntacic French lexicon, the Lefff .