[4M3-J-9] Natural language processing, information retrieval: inferring emotion and intension

[4M3-J-9-02] Automatic Indexing Kansei information based on Appraisal Dictionary From Reviews

〇Sho Hashimoto¹, Atsuhiro Yamada¹, Noriko Nagata¹ (1. Kwansei Gakuin university)

Keywords:Text Mining, Review Analysis, Appraisal Dictionary

In this research, in order to acquire knowledge that can be easily reflected directly in the product design, we propose a review analyse method which expresses the relationship between emotions, main impressions, and elements as indices. The method consist in 3 steps; (1) extracting topics of impression which express the affective difference of products with the appraisal dictionally, (2) indexing the relationship of these topics and emotions, and (3) indexing the relationship of these topics and component elements. Knowledge on how to change the elements to evoke a certain emotion is expressed as an index of the relationship between emotion - impression - elements. By interpreting this result, a knowledge such as "for evoking user's delight, it is important to give the impression of courtesy and quickness, and these impressions are improved by the shop's behavior when users present someone with a gift" can be obtained.