Data-Driven Contextual Valence Shifter Quantification for Multi-Theme Sentiment Analysis

Users often write reviews on different themes (e.g., an opinion aspect for restaurants or a movie category) and involving linguistic structures with complex sentiments. Considering the distinctions among themes, the sentiment polarity of a word can be different across themes. Moreover, contextual valence shifters, a family of linguistic patterns, may dramatically change sentiment polarity of contexts that they appear in. Both challenges cannot be modeled effectively in traditional sentiment analysis. Studying both phenomena requires multi-theme sentiment analysis at the word level, which is very interesting but significantly more challenging than overall polarity classification, due to inconsistency of word sentiment in different themes and sentiment shifting. To simultaneously resolve the "multi-theme" and "sentiment shifting" problems, we propose a data-driven sentiment analysis framework, MTSA, to enable both polarity predictions of the same word in reviews of different themes and, discovery and quantification of contextual valence shifters. MTSA formulates multi-theme sentiment by factorizing the review sentiments with theme/word embeddings and then derives the shifter effect learning problem as a logistic regression. To rigorously formulate the problem, a series of intuitive assumptions are proposed and later verified in extensive experiments conducted on real-world reviews. The improvement of sentiment polarity classification accuracy demonstrates not only the importance of "multi-theme" and "sentiment shifting", but also effectiveness of MTSA. Human evaluations and case studies further show the success of multi-theme word sentiment predictions and automatic effect quantification of contextual valence shifters.