Recently the desires to retrieve images intelligently based on higher-level information such as a user s intentions or sensibilities have been increasing, although there have been studies that a user associates sensibility words with given things or the lower-level information 1, for example, textile, color, and so on. As a way to satisfy the desires we suggested the KANSEI Vocabulary2 Scale in Sunkyoung Baek et als. (2006). The scale was intended to be a system that a computer might understand, retrieve, and recognize human beings sensibilities based on visual information including shape, color, texture, and pattern, and that through the recognition the computer might produce a resultant image desired by a user. For the scale we collected more than 200 adjective data from more than 250 people aging from 19 to 40 through survey.3 We couldn t use all of them, however, because of physical state description rather than feelings and difficult meaning with the data. Nevertheless we couldn t discard some of the data with only intuition. First of all, we needed a certain criterion to certify whether the surveyed adjectives were really sensibility words or not in a logical theory. In general most studies on English adjective have classified adjectives relying on word order used in phrases or their syntactical properties, while the adjective for KANSEI system can t be analyzed and chosen in accordance with a phrasal order or syntactical property since the property of the adjective is not syntactical, nor phrasal, but semantical. Therefore in this paper, we takes the focus on the adjectives used for the KANSEI Vocabulary Scale, and will examine some existing standards classifying adjectives and suggest a proper way to analyze and choose sensibility words.
1. Introduction
2. Adjective Classifications
3. Conclusion