Affective taste evaluation system using sound symbolic words

Abstract

Language and emotions play an important role in affective computing and advancement of artificial intelligence. Sound-symbolic words (SSWs) have become increasingly important for meaningful descriptions of perceptual experiences as a detailed and reliable vocabulary. In this research, we constructed a system for quantifying texture/taste impressions expressed by SSWs. This system decomposes input SSW into phoneme elements and refers to the category quantity of each phoneme element for each adjective scale referring to a quantitative rating database. Then, the system displays the evaluation values calculated by impression-rating predictive model. We anticipate that this system will be able to support food development and recommendation of food name which expresses the impression of goods.