Common Recommendation System For Commodities Based On Multi-algorithm Fusion

The rapid development of the Internet has brought new challenges to the recommendation system.The information processing of recommendation systems has changed from digging more information to digging more useful information.Information overload has become a focus of current recommendation system research.How to dig out effective information from excessive information is also an important direction of recommendation system research.The current commodity recommendation system is mainly carried out from the perspective of users and commodities.User-based recommendation is to actively collect users' historical behaviors and habit characteristics,sort them,count them,analyze them,and calculate them to generate personalized,accurate,and novel real-time recommendation services for each user,so as to dig out the user's potential Personalized recommendations for purchased products.The main problems of traditional recommendation algorithms can be summarized as follows:(1)Cold start and data sparse problems.Such problems are likely to occur in user-based personalized recommendations,because they are less aware of users and difficult to obtain new user information.(2)User interest change issues.People's hobbies may change with their age and experience,and users' interests will change with their needs.If this issue is not considered when building a user's interest model,it will be impossible to grasp the user's current interest and make accurate recommendations.This paper proposes a product common recommendation system based on multi-algorithm fusion.Most traditional product recommendation systems only consider the user's purchase record or the user's social relationship to improve the recommendation performance and solve the sparse matrix problem.From the perspective of users and commodities,this article proposes a commodity co-recommendation model.First,from the perspective of users,a TP-FIM algorithm based on data mining is proposed to filter data information,extract effective information,and narrow the scope of recommendation.On the other hand,considering the use of trust relationships in social networks,the use of random walk algorithm to determine the set of neighbors,by discovering the potential complementary relationships of products to improve the accuracy of recommendation and solve the problem of data sparseness and cold start of the recommendation system,and propose a method based on A fusion product recommendation algorithm based on social relationships.The superiority of the algorithm is analyzed and summarized,and the experiment proves that the quality of the recommended results meets expectations.Use the TP-FIM algorithm to filter product data,and calculate the user's preference for the product according to user behavior;Then combined with the neighbor set generated by the product fusion recommendation algorithm based on social network to realize common recommendation,and give the final recommendation list.In summary,the product common recommendation system based on multi-algorithm fusion involves a variety of key technologies.This article analyzes the problems of the traditional product recommendation system,and mainly improves it from the product recommendation model,social information filtering and social neighbor set generation.,And the experimental verification shows the superiority of improving the recommended results.