SENTIMENT RETRIEVAL ONWEB REVIEWS USING SPONTANEOUS NATURAL SPEECH

This paper addresses the problem of document retrieval based on sentiment polarity criteria. A query based on natural spon-taneous speech, expressing an opinion about a certain topic, is used to search a repository of documents containing favor-able or unfavorable opinions. The goal is to retrieve docu-ments whose opinions more closely resemble the one in the query. A semantic system based on speech transcripts is aug-mented with information from full-length text articles. Poste-rior probabilities extracted from the articles are used to regu-larize their transcription counterparts. This paper makes three important contributions. First, we introduce a framework for polarity analysis of sentiments that can accommodate com-binations of different modalities capable of dealing with the absence of any modality. Second, we show that it is possible to improve average precision on speech transcriptions ’ senti-ment retrieval by means of regularization. Third, we demon-strate the robustness of our approach by training regularizers on one dataset, while performing sentiment retrieval experi-ments, with substantial gains, on another dataset. Index Terms — Sentiment analysis, spontaneous speech reviews, polarity, subjectivity, information retrieval