Social Recommendation System Analysis Based On Agent Model

Under the trend of the rapid development of Internet technology, online shopping is more and more big, the proportion of the personalized recommendation system of also appears more and more important. Collaborative recommendation based on the user, however, recommend traditional way, such as generated user model does not show, but directly by comparing the similarity of the users, and on this basis to recommend, this recommendation effect is not prominent. Recommendation system based on goods to a certain extent, overcome a lot of the above shortcomings, but also give up the interest for the user is recommended. Given the recommendation of personalized recommendation system is low efficiency, and social recommendation system gradually become the research hotspot and focus.Social relations between social recommendation system mainly based on user information construction of the social relationship between user network, for a new user, as long as the network of a user with this user has a direct or indirect social relations, can according to this kind of social relations and the known user interest model, suitable to recommend to new users. Facts have proven that as people’s social attribute, social recommendation system recommendations will be more powerful.Based on the Agent model to study social recommendation system, content and objectives of the study is:(1) Based on Agent Model social networks. Social recommendation system is built up in the social network launched. Therefore, in the study of social recommendation system, we must first initialize a social network. Since independence Agent with encapsulation, so it can be packaged into social properties and parameters into Agent and Agent as a single node in the network, to build social network model based on Agent. In this section, there are two parts of the work:to design and build social networks Agent Model. In this paper, based on the characteristics of the consumer market, the main design of the two types of Agent, Agent namely producers and consumers Agent, as well as their modeling process and internal relations. On this basis, the use of dynamic and independence Agent, initiates a social network model.(2) On the basis of the social network model Agent based on the use of collaborative filtering algorithms work, build social recommendation system based on Agent. Recommended for the current system is mainly focused on the optimization algorithm recommended, recommended to optimize response time alone, while ignoring the recommended degree of personalization, recommendation or simply optimize the efficiency, but the lack of universality, this paper to study the social recommendation system by Agent model, Agent use powerful distributed computing capabilities, and is designed TopX algorithm and threshold threshold algorithm to find out the network model of the group with the highest similarity to the common characteristics of the group to make recommendations. The final experimental results verify the feasibility of the research ideas.