Image browsing for infomediaries

Information intermediary or infomediary is likely to claim a major segment of the revenue of e-commerce transactions. Infomediaries are e-commerce companies leveraging the Internet to unite buyers and suppliers in a single efficient virtual marketplace to facilitate the consummation of a transaction. Most of the search facilities in the infomediaries are text- and navigation-based, for example by keyword and category searches. The search facility can handle most of the products, but it is inefficient to handle feature-based products, such as textile and shoes with different colors and textures. It will be difficult for buyers to speck searches of feature-based products in describing the product features and characteristics. For example, it is hot easy to describe the pattern and color of a T-shirt using textural expression. In this paper, we propose an image browsing technique to facilitate search of feature-based products for infomediaries. We implemented an image-based retrieval system with the proposed image browsing technique for textile industrial and evaluated the performance of the browsing technique.