Research And Implementation Of Ontology-Based Profile Chinese Web Information Filtering System Prototype

Text filtering is a technology that can retrieve the text documents which satisfy the user's requirement in the dynamic documents stream. The application of the filtering has already been concerned in the field of web information processing, especially for the information collection and security based on content. With the rapid growth of the application of the text filtering, such as the e-mail, messege subscription and information security etc, the user's requirement for the performance of the filtering becoming higher and higher. In the task of the filtering, users pay more attention to the semantic information and even viewpoint information that they want to get, which is difficult for the traditional filtering techniques based on the statistics and the machine learning. In recent years, the research about semantic information filtering has become the hot focus, so we introduce the ontology into the information filtering to applying semantic information.Currently the web pages still carry the most information, and text is the main object of the web information processing. In this paper, we construct a prototype of filtering system to analyze and filtering the content of the web pagesThe filtering system constructed in this paper can be divided into two sub-systems --- one is for capturing and regrouping the data packets and the other is information filtering sub-system using ontology-based profile. This paper designs architecture to support various demands in the different environments. In the sub-system for capturing and regrouping data packets, we capture the data packets that pass the network card following the principle of the data packet sniffer, and analyze the data message at each layer of the TCP/IP, regroup the packets of the web pages, return the web pages. In the filtering sub-system using ontology-based profile, we introduce an approach to construct the user's profile based on ontology. Ontology provides a formal way to describe the semantics relations between the concepts by using the means of concept-properties model. Two algorithms have been designed to calculate the semantic similarity between feature vector and the profile, which have been impoved according to the evaluated results. The experimental result shows that the ontology-based profile for information filtering system can achieve encouraging result at the semantic filterings.