Research On Online Recommendation System Based On Trust Mechanism

With the development of the Internet,especially in the current era of the mobile Internet,individual lives are inseparable from the Internet.Individuals need to obtain useful information from massive amounts of information every day.Flooded by the large amount of online information,it becomes more difficult to sieve and obtain useful information for targeted audiences.In order to overcome this problem,recommendation systems are introduced into various Internet applications to recommend the content of interest to users.However,along with the ever-increasing demands for high-quality contents,and the requirements for recommendation systems constantly up the ante.Therefore,to get the desired effect,the research efforts on recommendation systems need to be prescribed in ever larger quantities,both in academia and in industry.Based on the current difficulties encountered by recommendation systems and the status quo of Internet applications,this paper exhibits an important practical significance to improve the experiences of recommendation systems by introducing trust mechanisms.This paper surveys the current research status of recommender systems worldwide,and carries out the following research work,the following work is carried out.(1)We introduce trust relationship data to build a trust matrix,which to some degree can alleviate the sparsity problem of the scoring matrix.Based on the characteristics of trust,we've studied the advantages and disadvantages of different ways in building the trust matrix.(2)For the online recommender system,we invented an incremental learning algorithm of matrix decomposition,with the help of some specific trust mechanisms.The experimental results show that the algorithm can greatly improve the training speed of the online recommendation algorithm with accuracy guarantee and provides a basis for quick update of the online recommender system.(3)This paper also presents an online recommender system based on the abovementioned incremental recommendation algorithm in the last section.Furthermore,we introduce a hot list of popular products,to combine with the recommended products after trustworthy configuration and ranking,which successfully alleviates the shortage problem of recommended products in our recommender system,meanwhile,the stability of the recommender system is accordingly improved.