R Study On An Ontology Based Sensitive Word Filter Method

Sited in XILINHOT, Xilin Gol Career College is a public vocational school, established in2003with nine vocational schools. Now, there are more than18,000students, nearly900faculty and staff in Xilin Gol Vocational College. With the rapid development of network technology, the college has to keep up with the wave of Internet information age. Therefore internet applications such as college student online forums, online office systems etc. are built through network platforms. However, meanwhile with the publication of information, there exists a lot of practical problems gradually with the evolution, such as the false advertising information, improper student speeches, sabotage and other reactionaries, etc. The integrated network platform has to face severe challenges. Facing the practical needs of Xilin Gol College network informatization, this thesis intends to design a sensitive word filtering method. Through the filter of advertising, improper speech, discordant words, it will help to provide flexible and accurate network information management.The main work of this thesis includes the following aspects.First, we provide a thorough study on the existing classical sensitive word filtering methods, and a comprehensive analysis with comparing the characteristics of existing methods, then found the fact that existing classical algorithms and extended algorithms put more emphasis on the efficiency of the filtration while ignore the accuracy. In response to this weakness, this thesis proposes a sensitive word filtering method based on ontology. The kernel idea is to classify the sensitive words, explore the possible semantic associations between the sensitive words and use them to narrow filter range, so as to improve the filtering accuracy.Secondly, in order to support the effective sensitive word filtering, we have established a college student forum sensitive word filter domain ontology for student information, sensitive word entries and management regulations. With the clarification of student properties, sensitive words semantic information and management regulations, we made the formal modeling of the domain. In the tool of Protégé, we exploited OWL language to describe the formal model. Then we build and will enrich an ontology with73classes,34properties and1,205individual instances.Again, according to the actual needs of college network information platform for the sensitive word filtering algorithm proposed, this thesis presents the design and implementation of the method. Through detailed analysis of aggregate demands and detailed demands, we present the overall and detailed design of the algorithm, and then using OWL‐API, Swing and some other java packages, the algorithm has been implemented.Finally, after system integration, we referred to the network information platform of the college to verify the feasibility and efficiency of the method, to verify the actual hard aspects to improve the accuracy of the method in terms of sensitive word filter. The successful completion of this thesis provides a new way of thinking for sensitive word filter. From the perspective of semantic associations in the range of sensitive words filter help to determine the searching range dynamically, and improve the accuracy of the goal. The proposed approach is another meaningful attempt of ontology engineering techniques in solving practical problems in the practical fields.