Complexity of lexical descriptions and its relevance to partial parsing

摘要: Complexity of Lexical Descriptions and its Relevance to Partial Parsing Srinivas Bangalore Supervisor Aravind K Joshi In this dissertation we have proposed novel methods for robust parsing that integrate the exibility linguistically motivated lexical descriptions with robustness statistical techniques Our thesis is computation linguistic structure can be localized if items are associated rich supertags impose complex con straints in a local context However increasing complexity makes number di erent each item much larger hence increases ambiguity parser This resolved by using supertag co occurrence statistics collected from parsed corpora We explored these ideas Lexicalized Tree Adjoining Grammar LTAG framework wherein disambiguation provides representation an almost parse used dis ambiguated sequence conjunction lightweight dependency analyzer compute noun groups verb linkages even partial parses shown trigram based supertagger achieves accuracy on Wall Street Journal WSJ texts Furthermore anal ysis output identi es links accurately exploited Explanation Based Learning improve e ciency approach limited domains modeled as Finite State Transduction implemented such system ATIS domain which improves factor