The Research On Personalized Recommendation Model Based On User Context In The Forum

Since the birth of the Forum, there have been many characteristics, such as richer content, stronger interactivity, higher field-independence and so on, which won the favour of a large number of network users of all ages and become one of the most popular information service systems. As a mainstream platform for information dissemination and interactive, forum has become one of the most important way that users could share their knowledge, exchange ideas and transfer information. The users surfing the BBS are relatively concentrated and have a long mean residence time, while there are also a large number of various new information and resources posted in the forum every day. Obviously, abundant information resources make it more convenient for users to obtain. However, people who want to find out their real favorite topic posts in a short time become more and more difficult."Information Overload" and "Information Trek" have become current urgent problems of Forum needed to solve.Traditional search engine systems could only provide correct retrieval results for those users who have definite needs, and in many cases, the needs of surfing users in the BBS is not very clear. Therefore, simply relying on search engine system to supply personalized information has been unable to meet individual personalized needs. Personalized recommendation system could recommend network users their interested information resources through predicting users’ preferences and interests. And, as for those who have no explicit needs, it also will generate a good effect in the aspect of finding their interested information resources. With the rapid development of the Internet, network users have put forward higher requirement for personalization. To improve network users’ utilization efficiency of forum posts resources and resolve the issues about "Information Overload" and "Information Trek", the research on personalized recommendation system of forum provides an effective method. Currently, some researchers have begun to study the user context introduced into personalized recommendation system. Forum has inherent advantages in the aspect of information resources and owns lots of user information. Therefore, the user context is introduced into the forum personalized recommendation system which can improve the accuracy and personalization of recommendation results to some extent.To promote the application of personalized recommendation system in the forums, this paper analyses the various existing personalized recommendations models and presents a Personalized Recommendation Model Based on User Context. This model optimizes the application performance of the Forum. The main works of this paper are as follows:(1) Through in-depth analysis on the existing personalized recommendation model, this paper considers BBS’s own resources and users interests and preferences, and proposed a Forum Personalized Reecommendation Model Based on User Context(PRUC) which considers users’context elements.(2) Extracting context information and constructing context conceptual models. There are two kinds of context information, such as user context information and topic posts context information, which is related to topic posts recommendation. User context information indicates the user’s current basic information, natural and social properties. Topic posts context information includes thread posts information and reply posts information.(3) The design of the key recommendation algorithms.Firstly, this paper designed User Context Clustering Algorithm to discovery neighbor users who have the similar interests and preferences with current users. Secondly, this paper designed User Context Extension Algorithm to expand users’own context information based on its neighbors’context information, aimed to find out current users’potential contextual information in the Forum. Finaly, this paper designed Sorting Algorithm Based on Contextual Inference Rules, through reasoning out the similarity between Extension user Context Ontology (ECO) and Topic Posts Context Ontology (TCO), and screened out the topic posts from the Form which are suitable to current users’interests and preferences.(4) According to current mature web development technologies, a personalized recommendation experimental Forum is designed and implemented based on actual operating data sets from a university Forum. The experiments are designed to verify the effectiveness of the model proposed in this paper. Experimental results show that the proposed method in this paper is applied to the personalized recommendation of forum topic posts. It has better characteristics of accuracy (includes precision and recall) and user satisfaction compared with other recommendation system model. This research on personalized recommendation system has a certain role in promoting personalized recommendation study, and explores the application effect in the Forum. And the research on personalized recommendation system of forum provides an effective method to solve the problems about "Information Overload" and "Information Trek". Therefore, the research has a certain practical significance both in the field of scientific research and practical applications.