| In the web 2.0 era, with the appearance of online communities, a large number of subjective texts expressed people's views, attitudes, feelings and emotion etc. have a rapid growth. The concept that "the network is the media" has been firmly established. Electronic word of mouth (eWOM) has caused increasing attention of researchers. Considering the consumer sentiment which we are concerned with, taking online product reviews as the main existing form, it is a vital information clue both to companies or potential consumers. Although the current text-based affective computing has made great progress, there are also deficiencies on the study of fuzzy essential attributes of natural language and human sentiment.In this paper, we focused on online product reviews, took the consumer muti-affective analysis as a main line, combining with the product object and behavior inference and using fuzzy sets and fuzzy logic as the mathematical basis for natural language. First of all, through the existing affective classification and the semantic features of online product reviews, the classification system of consumer multi-affective expression has been established, and on this basis the affective lexis fuzzy ontology database is built, in which the affective strength is described by fuzzy membership function. After corpus collection, an affective tagging system for online product reviews is also established based on the lexis ontology, through affective annotation, the muti-affective fuzzy corpus of online product reviews is set up. After then, the affective fuzzy computing is discussed. Based on the ambiguous language qualifier operators in fuzzy theory, a muti-affective fuzzy computing algorithm of online product reviews (MFCA-OPR) is introduced. This algorithm takes fuzzy computing for multi-product-attributes and multi-affective-expression of the online product reviews, which has some certain of innovation. Finally, the applications of multi-affective fuzzy computing are simply discussed. Based on Quadrifid Graph Model, the product competitive superiority is analyzed; and by constructing fuzzy inference rules based on consumer psychology and behavior, fuzzy recommendation for consumers is initially studied.The consumer affective computing and inference of online product reviews has important practical significance and bright application prospects in many ways. It also has a more important theoretical significance and application value for the current fledgling text sentiment analysis research, commercial information intelligence services and personalized recommendation technology in the emerging e-commerce etc. |
PDF Full Text Request