Research And Implementation Of Product Recommending Method Based On The Social Network

With the prosperity of some typical online social networking sites like Facebook, Twitterand so on, online social networks have become an important platform for the disseminationof news and views. The openness, richness in content of online social network platform anduser interaction information promotes the boom of online intelligent recommendationtechnology and virtual marketplace technology. Online social networking has affected thepurchase or use of the product or service. However, it has become a difficult problem forusers how they can get information they want from vast amounts of information for products.The existing e-commerce systems, such as Dangdang and Amazon can not take the initiativeto recommend accurate products to the users after the users enter these systems. Thesesystems can only give recommendations based on the users’ explicit or implicit preference,ignoring the ability for the social networks to recommend products. This paper aims atstudying a product recommending method based on social network, including the followingaspects:Firstly, this paper analyzes the researching status of domestic and foreign social networksand corresponding recommending methods. Then the existing problems in current productrecommending systems are given.Secondly, this paper puts forward a product description model SNPDM based on thesocial network. The model fully takes into account the structural characteristics ofsmall-world networks formed among users, provides richer information for the realization ofpersonalized product recommendation and reduces the searching space for subsequent productsearching.Thirdly, this paper proposes a product recommending method based on SNPDM which isable to resolve the poor searching quality problem and the problem of neglecting thecharacteristics of small world. The method includes: the similarity calculation methodbetween users and products, calculation method of users’ buying possibility, nearest neighborselecting calculation method, etc. This method, SPRM can be integrated into existing productrecommendation system to improve the accuracy and user satisfaction.Finally, in the design and implementation of experiments, the data from Dangdang andAmazon in China are used. The experiment results verify the correctness and feasibility of the proposed model and method.