Research On Judgement Methods Of Potential Friends Based On Behaviors

With the development of the Internet,the way to replace the traditional offline friends with online friends has been widely studied and applied.In addition,the progress of the Internet promotes the development of online shopping sites as more and more people are keen on online shopping.Shopping behaviors can reflect the contacts between users,which can be used to analyze the user's interest similarities.By analyzing user's shopping behaviors,this paper puts forward several groups of characteristics that represent the relationship between users,then conduct a study on the potential friend identification.By processing and analyzing the data of users 'purchase of goods,this paper puts forward the relationship characteristics of users,the similarity of product features,the recognition of product features,the similarity of users' emotions and the features of product selectivity.All these characteristics represent relationships between users.Moreover,the paper also comes up with the characteristics of comment usefulness and emotional strength to represent the characteristics of the user.Use the point-by-point method to calculate the final similarity between users,then build a "friend pool",and use the ternary closure theory to expand "friend pool" to increase the hit rate of potential friend recognition.Experiments show that it can achieve better recognition effect on the premise that the user's similarity is calculated by using the point-by-point method and then "friend pool" is constructed by this method.The introduction of ternary closure theory improves the hit rate of recognition results,and has great significance to the research of user recommendation.