Long-term relevance feedback and feature selection for adaptive content based image suggestion

Content-based image suggestion (CBIS) addresses the satisfaction of users long-term needs for “relevant” and “novel” images. In this paper, we present VCC-FMM, a flexible mixture model that clusters both images and users into separate groups. Then, we propose long-term relevance feedback to maintain accurate modeling of growing image collections and changing user long-term needs over time. Experiments on a real data set show merits of our approach in terms of image suggestion accuracy and efficiency.