Uncertain Multimedia Data Personalized Query System Design And Implementation

With the wide spread of multimedia technology, traditional uncertain data query andpersonalized information retrieval techniques can’t meet the needs of multimediaapplications. Therefore using the distributed data processing and personalized queryprocessing techniques to improve uncertain multimedia data query efficiency and thenimplement an uncertain multimedia data personalized query system is very important.Based on the analysis of functional and performance requirements, the architecture ofthe system is designed and the corresponding query model UG3M-UP is proposed. Themodel is based on the G3M Model, supports nesting and arbitrary uncertain granularity;uses the multimedia domain knowledge and hybrid feedback strategy to model userpreference; uses the SQL and XML hybrid storage strategy.Then the system that supports automatic acquisition and update of user preference andthe management of domain knowledge is implemented. The corresponding personalizedquerying algorithms that used in centralized and distributed environments are designedand implemented respectively. The centralized algorithms use the R Tree index and bestfirst strategy, whereas the distributed algorithms uses the divide and conquer method, andcan adapt to different network conditions.The experimental results show that system algorithms have higher recall and precisionrate than traditional database queries. Compared with modern skyline algorithms, thecentralized algorithms have better performance, and the distributed algorithms are moreefficient in mass data environments. Proving that the system is capable of meeting users’needs of personalized querying on uncertain multimedia data.