Study On Recommendation Method In Social Commerce Based On Hypergraph Model

With the rapid development of e-commerce and social network,social commerce is gradually emerging.Especially with the popularity of the "Internet+",traditional ecommerce is also gradually upgrading and transforming and integrating into more and more social elements.Social commerce has changed the information one-way communication model of traditional e-commerce,increasing the opportunity for users to interact each other online,and improved the users' stickiness and users' conversion rate.However,with the arrival of the era of big data,the information overload caused by the problem is also increasing,how to recommend commodities to social commerce users among the magnanimous information to meet the needs of them to the greatest extent is imperative.Accurate and effective recommendation method has become one of the key technologies to attract potential users,reduce users' blind search time and convert customers' purchase rate.Designing a reasonable recommendation method of social commerce is an essential prerequisite of the social commerce recommendation.But the research on the recommendation method of social commence is still staying at the traditional e-commerce recommendation level,social commence involves more Information of multi-source heterogeneous objects,traditional recommendation methods have been unable to solve the problem of recommendation in the social commerce effectively.Therefore,it is necessary to discuss the recommendation method applying to social commerce.According to the characteristics of social commerce and the characteristics of hypergraph model,which can build a model about multi-type data nodes,we think that the process of analyzing multi-types of information about users and products in the process of social commerce recommendation can make use of hypergraph model to tap the target user's interest preferences fully and capture the needs of the target user's information accurately.In the social commerce scenario,because of the massive data,heterogeneous types and the problem of user cold start in the process of recommendation,we first set up the hypergraph model of the multi-type object information in the social commerce scenario,then cluster the multi-type object information,and then in each cluster rank each item that the target users interested in.According to the ranking results,we can recommend the former products in the top to the target user to complete the recommended process in social commerce.At present,some research fields related to recommendation methods have attracted the attention of many experts and scholars,and have achieved quite mature results.However,considering the attributes of society and big data of social commerce,the massive data that generate rapidly show multi-source heterogeneous characteristics the research method about recommendation method of social commerce for multi type information is seldom involved,and a systematic research system has not been formed yet.While the hypergraph model is able to put different types of data objects in a unified framework for research.Therefore,it is necessary to study the recommendation method of social commerce based on hypergraph,which is of great theoretical and practical significance.This paper mainly completed the following research work:(1)In order to solve the high dimensional and complex relationship among multitypes of objects in social commerce,construction method of hypergraph model of social commerce is proposed by the study of hypergraph theory.(2)Through the analysis of the hypergraph model of the social commerce,in order to alleviate the problem of the cold start in users recommendation process effectively and shorten the time of online recommendation,according to the big data characteristics of social commerce,we proposed a clustering method of social commerce based on hypergraph,so that it can be better applied to the social commerce recommendation which contains multi-type data objects.(3)On the basis of the results of clustering of social commerce based on hypergraph model,social commerce ranking method based on the result of clustering is proposed.The method can run in parallel in each hypergraph cluster,which can effectively sort the products that users are interest in.In the ranking,the former product is recommended to the corresponding target users,at the same time the social commerce recommendation of products is completed.(4)Combining with the actual social commence website,this paper puts forward the recommendation method of social commence based on hypergraph in the process of recommendation of the commodity in the social commerce scenario,and according to the characteristics of the actual website and the collected data,we build the reasonable model and recommend the products that users are interested in.Finally,we compare the results of recommending effect evaluation with the classic recommendation to prove the feasibility and effectiveness of the proposed method.The result shows that the proposed recommendation method of the social commerce based on hypergraph model can provide reference for commodity recommendation in social commerce,and the clustering algorithm also provides a reference for the clustering problem of multi-attribute data information.In addition,this method can also be extended to recommendation of friends or labels recommendation and other fields in social commerce scenarios,which has a high reference value.