The Research Of Personalized Recommendation Methods Based On Time-sensitive For E-commerce

With the increasing of recommendation algorithms in e-commerce applications,the benefit of website is increased.As an important factor,time factor affects recommendation algorithms,so that cannot be ignored.Therefore,this paper researches recommendation algorithms based on time-sensitive applications,and applies it into e-commerce.In addition,this paper verifies the feasibility of the proposed algorithms through experiments.The main work of this paper is shown as follows:1.This paper researches the influence of the relative order of users' behaviors to users' similarities.Through the longest common subsequence algorithm,this paper finds the longest sequence under the relative order of users' behaviors.In addition,this paper computes the contribution degree of the list to users interests through extracting the non public interest list based on time-sensitive between users.Furthermore,this paper combines the contribution degree with UserCF algorithm by weighting.Through comparison experiments with other recommendation algorithms on two data sets,this paper verifies the good performance of the proposed recommendation algorithm on recommendation accuracy and recall.2.This paper processes the records of users' behaviors through Hash,and decays the weights of every record in accordance with the time in order to simulate the dynamical changes of users' interests,thus forms the fingerprint algorithm of users' behaviors.Through comparison experiments with other algorithms on two data sets,this paper verifies the good performance of the proposed algorithm on recommendation accuracy and recall.3.When computing the similarity between resources,this paper considers the attributes and popularity of resource with time-sensitive into substance diffusion algorithm.Through comparison experiments with other substance diffusion algorithms on two data sets,this paper verifies the good performance of the proposed substance diffusion algorithm with time-sensitive on recommendation accuracy and recall.In addition,this paper combines the above algorithm with ITEMCF algorithm with time-sensitive.Through comparison experiments with substance diffusion algorithm and ITEMCF algorithm on two data sets,this paper verifies the good performance of the proposed algorithm on recommendation accuracy.