Research Of Context-aware Recommendation Method Based On Rough Set Attribute Reduction And Collaborative Filtering

In recent years,with the rapid development of information technology,especially the Internet technology,Internet thinking has penetrated into all walks of life at an alarming rate,which resulting in a variety of Internet applications and massive data information.These vast amounts of information provide a rich source for people to obtain information content,but also cause the problem of “information overload”,which brings huge information burden to people.How to obtain information that satisfies their own needs from a large number of unrelated or redundant information in a timely and accurate manner is an urgent and prominent issue.Although the information retrieval technology represented by search engines has satisfied the certain needs of people,it can't satisfy the personalized query request of different backgrounds,different purposes and different periods because of its general nature.Personalized service technology is proposed for this issue,which providing differentiated services and content according to the difference in service information requirements of each user.The recommendation system,as an important branch of the research field of personalized service system,provides an effective solution to the problem of “information overload”,and has made great progress in practical applications such as e-commerce(Amazon,Netflix,etc.),information retrieval(Grouplens,Baidu and so on),as well as mobile applications,tourism and other practical applications.The concept of Context-Aware Recommender Systems(CARS)was first proposed by Adomavicius and Tuzhilin et al.and they believe that considering the user's context information in the recommendation system can improve user satisfaction and recommendation accuracy.Most of the existing context-aware recommendation systems assume that different context information has the same weight in the recommendation process;however,in actual applications,there is a difference in the effect of each contextual information on the overall system.How to calculate the influence degree of these contextual information to the recommendation result,and to add these influence factors to the calculation process of the recommendation system to improve the recommendation quality and accuracy of the recommendation system is a subject of great research significance and practical value.For this issue,the work done in this paper is as follows:1.Summarize the traditional recommendation system,study the theory and method of context-aware recommendation system,and classify and contrast various context-aware recommendation generation techniques.2.Review the basic concepts such as equivalence class,indiscernibility relation,and positive region of rough set theory,introduce the digital features such as attribute importance and knowledge dependence,and finally introduce the concepts of reduction and kernel,analyze knowledge reduction and attribute reduction algorithms for decision tables.3.An efficient attribute reduction algorithm for rough set is proposed.The attribute reduction algorithm is used to reduce the conditional attributes in the context information system.A context-aware user similarity calculation method is designed,and collaborative filtering is used to generate the attribute reduction algorithm.It is recommended to build a context-aware recommendation algorithm based on rough set attribute reduction and collaborative filtering.4.Compare the proposed algorithm with the existing context-aware recommendation algorithm and the traditional collaborative filtering algorithm,and compared with the simulation experiment.Experimental results show that the proposed context-aware recommendation algorithm based on rough set attribute reduction and collaborative filtering has significant effect on improving the quality and accuracy of the recommendation,and verifies the feasibility and efficiency of the proposed algorithm.