Agent-based Information Integration And Implementation Of Information Filtering Mechanism

With popularization of DTV (Digital Television) , DTV-oriented intelligent information service system gradually becomes the focus of research. The system aims at providing both individualized and popular service. Many subjects related to this system deserve to study.Based on the project, the key technologies of the system including information filtering technology and information merging technology are studied in this paper. The architecture of the system is designed and the corresponding algorithm is given as well. At last prototype of the system is realized with Agent technology. The main content of the paper is as follows:Firstly, the model, method and corresponding theory of information filtering are studied. Information filtering is realized by using VSM (Vector Space Model) according to the characteristic of DTV-oriented intelligent information service system. Negtive feedback threshold-obtaining method is brought forward and the feedback learning formulation is established.Secondly, the policy, method and and corresponding theory of information merging are studied. Multiple user preferences information merging algorithm based on total distance minimization is designed according to belief revision theory. The algorithm satisfied IC (Integrity Constraints) postulates and Majority postulate in belief revision theory. Because the calculation result using the traditional symmetry-distance is not fit to human being's common mentalities, the new concept of asymmetry-distance is brought forward. The better merging result is achieved which is fit to human being's common mentalities with the concept of asymmetry-distance applied in merging process.Lastly, the design goal and architecture of the DTV-oriented intelligent information service system are put forward. The prototype system is realized with Agent technology applied. The system is efficient and easy to extend and transplant.