SUBJECTIVITY WORD SENSE DISAMBIGUATION: A METHOD FOR SENSE-AWARE SUBJECTIVITY ANALYSIS

Subjectivity lexicons have been invaluable resources in subjectivity analysis and their creation has been an important topic. Many systems rely on these lexicons. For any subjectivity analysis system, which relies on a subjectivity lexicon, subjectivity sense ambiguity is a serious problem. Such systems will be misled by the presence of subjectivity clues used with objective senses called false hits. We believe that any type of subjectivity analysis system relying on lexicons will benefit from a sense-aware approach. We think sense-aware subjectivity analysis has been neglected mostly because of the concerns related to word sense disambiguation (WSD), the problem of automatically determining which sense of a word is activated by the use of the word in a particular context according to a sense-inventory. Although WSD is the perfect tool for sense-aware classification, trust in traditional fine-grained WSD as an enabling technology is not high due to previous mostly unsuccessful results. In this thesis, we investigate feasible and practical methods to avoid these false hits via sense-aware analysis. We define a new coarse-grained WSD task capturing the right semantic granularity specific to subjectivity analysis.