Application And Research Of Random Walk Model In Personalized Recommendation Technology

With the continuous development of e-commerce and the explosive increase of network information,there is a growing possibility for citizens to be confused by the information.On the one hand,there is more and more information for people to choose,on the other hand,it is difficult to extract what their real need from the massive information.Although,to a certain extent,the information retrieval can alleviate the problem of information overload,it can not provide targeted and personalized service to the user's interest and,behavior.In this context,the personalized recommendation algorithms come into being.In the paper,on the basis of conventional recommendation,the model of random walk based on bipartite network is built.We put forward a similarity measurement based on implicit feedback.In this method,a uneven character vector is imported(the weight of item in the system).We put forward a improved random walk pattern which makes use of partial or incomplete neighbor information to produce recommendation information.In the end,the experiments on the real data set prove that,the recommendation accuracy and practicality are improved.Result of the experiment are true and reliable.Among all recommendation algorithms,collaborative filtering algorithm(CF)is he most widely used and the best effect.However,CF can not provide a good solution to the problem of sparse data and cold start.In the paper,the trust relationship between users is adopted to simulate the social relationship among the users.The trust network is chosed to replace the similar neighborhood.But,the user's data is sparse,which makes us pay more attention to the trust relationship between users,even the indirect neighborhood of the weak trust relationship,and it may lead to the reduction of the accuracy of the recommendation.Therefore,we propose a random walk model combining the trust relationship,score and the random walk model proposed which,can effectively solve the cold start problem and the low recommendation precision.Finally,on the Epinions data set,the proposed algorithm is compared with the traditional collaborative filtering algorithms and trust-based recommendation method,and result prove that our algorithm obtains better results.