A More Effective Book Search System Utilizing Colors Reflecting Reader Impressions

Shota Nakamura, Tomoko Kajiyama

Abstract: A search system is presented that supports hook search by users with ambiguous information needs. Readers' impressions and feelings about a book are extracted from reviews of the book, colors are extracted using relationships between feelings and colors, and the extracted colors are applied to the spine and cover images. The system is designed for multi-faceted search, and the retrieved results are displayed on a virtual bookshelf. It was evaluated using a database of 500 books. The books were categorized, the colors of their spines were decided on the basis of the extracted colors, units of 100 books in database registration order as the search facets. Testing with 12 participants revealed that the system facilitates searching for and finding books on the basis of ambiguous information needs quickly and accurately and thereby provide users with higher search satisfaction.