Improvement Information Filtering Model Based On The Knowledge Representation

With the variety of information, the biggest problem is not the absence or shortage of information, but the soaring of information quantity, which has already surpassed the processing capacity of people. How to provide a personalized information service for different information requirement of different users, using the same increasing information resource, is becoming a more and more important issue in the research field of information.The objective of personalized information service is to meet a special information requirement of a particular user. On one hand, the abundant information resources provide plentiful relevant information. On the other hand, a great deal of redundant information obstructs the acquirement of useful information. The information requirements of users are transited from the complete ownership of information resource to acquiring the valuable information.In this paper, an improved information-filtering model of personalized recommending technology oriented has been proposed. Firstly, a formal definition and framework of the model has been given in the paper, which lays a solid theoretic foundation for further study. Secondly, in the paper a user's requirement model has been proposed using the ontology theory that presented by descriptive logic, the model adequately expresses the semantic information of users. Next, the interminable original text forms a set of the meaningful paragraphs, and then the set of paragraphs is transformed into the ontology forest. Finally, a similitude comparison algorithm based on tree structure has been designed, which compare the requirement ontology to text ontology in semantic level of similitude.A personalized recommending system of electronic commerce has been designed on the improved information filtering model. The semantic factor in the expression of users' requirements has been taken into account in the system, which enhances the precision and the recall greatly and accomplish the recommendation function better. At last, a prototype of personalized recommending system of electronic commerce based on the improved model has been accomplished in the paper, and the prototype has been evaluated in the paper.