A Fuzzy-Based Multimedia Content Retrieval Method Using Mood Tags and Their Synonyms in Social Networks

Abstract

The preferences of Web information purchasers are rapidly evolving. Cost-effectiveness is now becoming less regarded than cost-satisfaction, which emphasizes the purchaser's psychological satisfaction. One method to improve a user's cost-satisfaction in multimedia content retrieval is to utilize the mood inherent in multimedia items. An example of applications using this method is SNS (Social Network Services), which is based on folksonomy, but its applications encounter problems due to synonyms. In order to solve the problem of synonyms in our previous study, the mood of multimedia content is represented with arousal and valence (AV) in Thayer's two-dimensional model as its internal tag. Although some problems of synonyms could now be solved, the retrieval performance of the previous study was less than that of a keyword-based method. In this paper, a new method that can solve the synonym problem is proposed, while simultaneously maintaining the same performance as the keyword-based approach. In the proposed method, a mood of multimedia content is represented with a fuzzy set of 12 moods of the Thayer model. For the analysis, the proposed method is compared with two methods, one based on AV value and the other based on keyword. The analysis results demonstrate that the proposed method is superior to the two methods.