Research And Implement Of Recommendation Algorithm Based On Social Network Influence

The population and online time of netizens increase with the development of Internet inrecent years, especially the development of intelligent mobile device and mobile Internet.With the trend, the social relationship is removing to the network from the reality, and finallythe social network was born. People with the same interests get together in the network, andfocus on the information they were interested. Then, the Internet aggregates more and moreusers. In this aspect, many network companies perform good examples such as Alibaba, Sinamicroblog and Facebook. The system that recommends the suitable goods and advertisementfor the potential customers according to the data mining on their interests is calledrecommendation system.In this paper, the recent study on social network and recommendation system are firstlyintroduced. Then, a data modeling is formed for the social network. Besides, the concepts ofusers, user relationship, item characteristic, user preference and algorithm are also carried outto create a social network map. Then, a query method based on the map is put forward，and acalculation method combined random forest classifier is formed.With the calculation method, a model based on B/S is developed, which contains aportable and modularitied recommendation system. In the paper, a general introduction for thesystem will be gave out in order to provide the details of the main modules. Then, anillustration for development environment and relative technology will be introduced. Ademonstration for the system is also performed.All in all, the calculation method put out from DS is used to perform the parameters test,cold boot test and algorithm time-consuming test. According to the experiment analysis, theproblem of cold-start can be resolved based on the calculation method on the social network,and a good preferment for an effective calculation is created because of the cache policy.