Representation of user preferences and adaptation to context in multimedia content – based retrieval

The task of content – based retrieval is to provide users with the multimedia documents that best match their wishes. This process is not free of uncertainty; the role of the user profile is to remove a part of this uncertainty, using the information it contains concerning the user’s preferences, and thus improve the precision. For this aim MPEG-7 introduced description schemes representing users’ preferences and tools supporting user interaction. In this paper, using a formal representation of user preferences and a semantic knowledge base, we expand the user profile representation introduced in MPEG-7 in order to acquire more semantic information about the user. Furthermore, using the notion of context, we identify the part of the profile that is related to the user’s query, thus preventing irrelevant interests from affecting the process of content – based retrieval.