Design discovery for social recommendation of Web graphics
J. Tatemrua, Kitaru Suzuki · 2002
To apply social recommendation to image databases, we have developed an clustering algorithm that takes account of both social and content based similarity between image items. Resulting clusters are called "design groups" since it represents visual features appealing to users. The system organizes image items and recommends designs that will appeal to the user. We have applied this technique to a Web graphics database and evaluated its effectiveness by user testing.