Modeling and retrieval of multimedia personalization
|School||Huazhong University of Science and Technology|
|Course||Computer Software and Theory|
|Keywords||Multimedia personalized model Domain knowledge Hybrid feedback Semantic Query Queries based on domain knowledge|
As computer networks and the rapid development of multimedia information , the user's search needs are constantly changing , users are no longer satisfied with the text retrieval , multimedia information retrieval concern. Currently available multimedia personalized models are mostly based on modeling user search terms , without taking into account semantic information and multimedia categories , can not meet the different users of the individual needs of multimedia information . Therefore, the design of a support multimedia semantic information and taking into account the individual needs of the user retrieval methods , has important significance. In G3M the common MPEG-7 descriptions stored , based on the domain knowledge into G3M model, to achieve the description preferences of individual users , create a multimedia model for personalized G3M-UP. The model description files from MPEG-7 and introduced to extract semantic domain knowledge to achieve user personalization preferences description , make full use of information and to extract semantic description of user personalization preferences of individual users to achieve the sort query results , using the extended G3M storage policy preferences of individual users to save and query history , with user query history and preferences of individual users describe the formation of mixed user feedback information to achieve the preferences of individual users to update the information . Comprehensive analysis G3M storage advantages, G3M storage infrastructure extends the SQL XML hybrid storage solution : take full advantage of the characteristics of domain knowledge , G3M-UP model describes preferences of individual users and user feedback composition . Combined with the diversity of multimedia queries and G3M-UP models , this paper proposes a variety of query options, including : semantic queries, full-text search queries , domain knowledge queries, create multimedia information query . Experimental results show that the query query scheme with high efficiency to satisfy the user query multimedia individual needs.