Profile Generation from TV Watching Behavior Using Sentiment Analysis

Abstract

This paper proposes a method for generating user profile from user's TV watching behavior using sentiment analysis. Personalized technologies such as information recommendation are currently hot topic of Web intelligence. Among them, TV program recommendation is expected to be one of the practical applications in near future, as digital TV service and partner robots providing personalized support are getting into our living environment. The proposed method does not estimate user's interest in a TV program only from its watching time as most of existing methods do, but also from user's utterances by applying sentiment analysis. The method employs fuzzy inference for estimating the user's rating of a TV program. A user profiles with bookmark format is generated based on the estimated rating of TV programs. Experiments are performed by collecting TV watching logs with diary-based approach, and the results show the proposed method can generate a user profile that can reflect user's interests.

Opinion Mining Using Fuzzy String Searching

“…Subrahmanian and Reforgiato graded sentiments by the combination of adjective, verb and adverb [11]. In contrast to the algorithms that extracted the sentiments using adjective -verb combination or adverb -adjective combination, the model was trained using adjective, verb and adverb combination.…”

Personalized TV-program recommendations based on life log

“…However, it has become difficult for users to find interesting TV channels or programs from among the multitude of channels and programs available. Various recommendation systems, which estimate user preference for TV programs based on past TV watching histories, have been researched with the goal of recommending TV programs [1][2][3]. However, since only very similar programs tend to be recommended by these systems, the user will rapidly lose interest in the recommendations.…”