Research And Application On Knowledge Personalized Recommendation Technology Based On Knowledge Scenario

With the spread and applications of information technology, many enterprises have stored the information about customer, product development and collaboration knowledge in their internal information management system. Meanwhile, a large number of information(mainly expressed in the Web page) which enterprises interested in also appeared on the internet. The combination of internal knowledge and external knowledge can help people to derive useful knowledge, increase the enterprises' technology, competition, business and strategic intelligence. Because of its intiative and timeliness, knowledge personalized recommendation have became a major development direction in the application of knowledge. However, there are many limitations about application of knowledge, such as knowledge difficult to understand, new knowledge difficult to obtain, knowledge overload, these shortcomings significantly limit the development of knowledge application. As knowledge is only have its value in a particular environment, it is necessary to related knowledge and knowledge scenario-conditions, background and environment about knowledge and knowledge activities-to solve these problems and difficult in the knowledge application.In this paper, personalized recommendations demand in ITV pretreatment system of telecommunications business is as application background. The paper proposes personalized recommendation method based on knowledge context, describes the system framework and operational resources of personalized service, and researches the acquisition of user interest, the model of user historical context and its association with knowledge, personalized program push based on individual situations, and so on.The main findings are as follows:1. In the demand of ITV programming personalized recommendation, This paper proposes personalized recommendation system's architecture based on knowledge context, designs and achieves the various functional modules, and implements the combination of knowledge context and personalized recommendation. 2. By analysis the programming information and use the Ontology model, the unified programming context model has established.3. Using log files and associated analysis tools, we analyzed user's behavior feature and collected user basic information and historical operation records. Based on programming context, we constructed a unified situation model for user interest by using ontology model and stored the situation model as user interest model.4. The method in this paper by calculating the similarity of user interest model identify the current user's similar users and by using collaborative filtering technology recommend the similar users favorite programs to the current user. This method solved the user score parse problem in general collaborative filtering methods.5. The programs personalized recommendation algorithm based on customer interest model is applied to telecommunications ITV pretreatment system, to achieve a combination of theory and application.