Using Context to Understand User Intentions in Image Retrieval

This paper discusses how combining existing techniques in a new way may help improve the understanding of user intentions in image retrieval. This is an especially challenging task in situations where the information need of a user is ambiguous, or if the user is unable to express it. The suggestion made here is that combining two different approaches to image retrieval with the utilization of context information and user interaction will help alleviate this problem. This paper describes how an approach that uses visual data, textual data in the form of context information, and user interaction in the image retrieval process may be developed and evaluated.