Research On Recommendation Model And Application For Mobile Commerce Based On User Context Interest In Cloud Environment

With the explosive growth of mobile Internet, rapid development of electronicbusiness and rapid popularization of intelligent mobile phone, a mobile Internet newecology is emerging globally. But the particularity of mobile commerce make thetraditional recommendation system diffcult to meet the special needs of this digitaluniverse era. The purpose of this study is to combination the user situational interest andcloud computing technology to propose a cloud-oriented users situational interestrecommendation model, thus solving the trust recommendation, multi interestrecommendation and service of quality recommendation. According to the characteristicsof this paper, we use collaborative filtering method, social network method, ant colonyclustering method and ant colony neural network method.The contents in this paper are listed as follows.Firstly, aiming to resolve the mobile commerce scenario suggested problems, cloudcomputing technology and mobile user context are combined to propose a collaborativefiltering model based on user interest in mobile scenarios. Through computing the scenesimilarity based on mobile users, we find similar scenarios constructed target user setcurrent situation, and then establish the project scoring scene and scoring matrix. Basedon MapReduce, we propose a collaborative filtering recommendation method to realizeparallel recommendation.Secondly, to address the traditional collaborative filtering of data sparsity and coldstart problem, a cloud user situational interest recommendation model under differenttrust information environment is proposed. A mobile commerce situational interest andrich trust information recommendation model trust relationship is introduced to solve thedata sparseness problem existing in collaborative filtering algorithm, and processingmethods to solve complex social network recommended by MapReduce data is proposed.Mobile commerce recommendation model of situational interest and sparse trustinformation is mainly devoted to solve trust less information available the realitycircumstances lead to inaccurate problem based on the recommendations, specifically the situational interest similarity matrix and potential trust degree matrix are combined into acomposite matrix. Then, we use the Resnick recommended formula and MapReduce dataprocessing method to implement the recommender model.Thirdly, to address the user single interest problems, a representation model is givenbased on mobile user multi interest level. At the same time, ant colony algorithm andhierarchical clustering method is used in the mobile commerce ant colony clusteringprocess. The level of user situational interest is used in the target function to generateclustering hierarchical clustering algorithm and the new system tree, and constructmulti-level ant colony search path is found according to some target user’s nearestneighbor cluster. Then, we use other users within the cluster on the target score to predictproject the not-scored items. Finally, the combination of MapReduce and collaborativefiltering recommendation algorithm is designed in the experiment.Fourthly, to resolve the service quality of mobile commerce preference predictiverecommendation problem, a hybrid recommendation model is presented for theprediction of user location scenarios. Firstly, based on the user context information tocluster all location service and users of the site by the autonomous system, we build auser-service matrix; and then, the user and the forecast of project method based onpredictive value is given to train the weights of Ant Colony Neural Network based onMapReduce. Finally, according to these weights, the system gives the final servicequality prediction value.Finally, to implement the model in real world, a mobile commerce siterecommendation framework is designed in the empirical research. First of all, weestablish a mobile commerce tourism attractions recommendation system framework. Onthe basis of constructing the model of user oriented situational interest, we make anempirical study of students in Qinhuangdao university. Empirical results indicates thefeasibility of this scenic spot recommendation system to meet the personalized needs ofmobile users.