Research And Implementation Of Trust-based Recommendation System In Big Data

Nowadays,with the rapid development of the Internet,a large amount of information is also produced.Although it provides more value for people,it also means that it is more and more difficult for users to quickly and effectively collect and obtain information that meets their own needs.The recommendation system,as a commonly used and effective method and means for filtering data information,has received more and more attention from people.Although the collaborative filtering algorithm is easy to understand,and only depends on the characteristics of the user's past historical score data.But it has the disadvantages of cold start,sparse data and difficult to scale.At the same time,with the rapid development of new technologies such as cloud computing,the current data volume level has even reached the ZB level.The large amount of data generated affects the performance of the system in a stand-alone environment.In order to make the recommendation system obtain better recommendation results in the big data environment,this article has carried out relevant research on the recommendation system and the key technologies in big data processing.The main research work and innovations of this article are as follows:Aiming at the problems of cold start and sparse data,a model for trust transmission on the network is proposed.First of all,based on the direct trust relationship that consumers have,use the constructed model to predict the indirect trust relationship between consumers,and effectively solve the problem of sparse data.Second,on the basis of the trust propagation model,a random walk algorithm is used to build a trust-based collaborative filtering recommendation model,which improves the calculation of trust between consumers and distinguishes trust between consumers.Give higher weight to consumers with high trust.Finally,the relevant data set of EPN is used for relevant experimental analysis and design to verify whether the improved model proposed in this paper is efficient.It can be seen from the experimental results that the improved model can better improve the range of recommendations to users and alleviate problems such as sparse data.Compared with the Trust Walker recommended model,the model achieves better MAE and PS indicators.In view of the scalability and computational efficiency of the improved recommendation model in the face of big data in the context of big data,the model is proposed for parallel analysis and processing on the Map Reduce processing platform.At the same time,according to the mode corresponding to Key-Value,according to Key reordering,a large problem is divided into sub-problems to solve in parallel,and then the value with the same key is processed through the Reduce function.The function of the Map function is to divide the problem into sub-problems.The problem is then solved.The Reduce function merges the sub-problems after division and solution,and finally obtains a result.The results obtained were used in comparative experiments.From the experimental results,it can be seen that after the parallel operation,the model solves the problem of cold start,reduces data sparsity and improves the accuracy of recommendation,while the scalability and calculation efficiency also show good results.In short,it has conducted in-depth research on the key technologies of the recommendation system and big data processing,and proposed a method of building a trust model and parallelizing the problems in the recommendation system under the big data environment.The results of this study provide ideas for the research of algorithms and technologies in the recommendation system under the environment of big data,and have certain reference significance.