A User Recommendation Algorithm For Social Ecommerce

With the permeation of the social media and e-commerce,the social commerce,which is a new business model,is developing rapidly.Based on recommendation?sharing and word-of-mouth,the social commerce is linking the crowd of social and online shopping together cleverly.The social commerce will lead the developing of the e-commerce in web2.0.Many studies have shown that,the spread of word-of-mouth and friends'recommendation are important factors that affect users purchase decisions.Besides,interactions between friends can also produce new consumption needs.Social commerce is focus on this point,by building shopping communities,social media and other forms of user network cluster,and to promote consumption.Unlike traditional e-commerce,social commerce pay a lot of attention to the influence that factor bring to users' purchase decision such as sharing,recommendation and interactions between users.As a result,its user recommendation model is also different from the traditional e-commerce.This article is focus on the characteristics of social commerce and aims to put forward a new user recommendation model that suitable for social commerce.Firstly,with social network analysis method,we firstly studied the characteristics of the social network in social commerce and explored the specific influence that social network structure bring to users' adoption behavior.The results showed that the social network structure in social commerce meet the small-world and scale-free features of complex networks and the collectivization and modularity of the social network have a strong positive correlation to users' adoption behavior.Based on the communication to persuade theory,we secondly selected the potential factors that may affect users' assumption decisions from the presenter,the information receiver,the information itself and the relationship between the presenter and receiver.We found that the preference similarity,professional ability and social relations are core elements that influence users' decisions.According the study above,then we built the quantitative model of the three factors and proposed the recommendation model for social commerce finally.Using the real data from Epinions.com,we compared it with other five systems to evaluate its performance.The experimental results shows that the new recommender systems is quite promising in terms of mean absolute error(MAE),prediction precision,and recommendation precision compared to the traditional ones.At the end of the article,we proposed some suggestions to the social commerce companies according to this study.