Research On Points Of Interest Recommendation Algorithms Adapting To Context Changes

Collaborative recommendation is the most studied and widely applied in recommendation system.The problem of data sparsity is its inherent characteristic,which is more obvious in the recommendation of points of interest.Mitigating data sparsity is an urgent problem to be solved in recommendation.In addition,many context-aware recommendation systems take the context as a feature to participated in user modeling and the context will obtain a certain value.That ignores the problem that the impact of some contexts on user will change under different context combinations.Therefore,an effective method is needed for this case to do further analysis.In order to solve the above problems,this paper proposes Points of Interest Recommendation Algorithms Adapting to Context Changes.The algorithm models user preferences from the item labels,and uses contextual utility differences for pre-filtering collaborative recommendation.This paper was funded by the Science and Technology Research Project of Jiangxi Provincial Department of Education,No.GJJ170211-Recommendation method of Location Correlation Information for Adapting to situational changes.The specific work is as follows:(1)Introduce a labeling system,use-label to represent project features,and construct a user label scoring matrix to describe user preferences,which can effectively alleviate matrix sparsity and reduce matrix dimensions.(2)The user feature vector is represented by the user's score on the label.This vector is used to perform user clustering and user similarity calculations.And then the ratings were predicted.The user-label scoring matrix is a dense matrix,and each user will search for neighbor in the whole matrix.User clustering can effectively reduce the search range and improve the search efficiency.Experiment show that Improved Similarity And User Clustering Collaborative Filtering is superior to traditional collaborative recommendation.(3)Distinguish between labels and contexts,put forward the concept of contextual utility,and use the contextual utility difference to perform context pre-filtering.Many context-aware recommendation algorithms ignore the problem that the same context has different influence on users under different context combinations.This paper proposes Points of Interest Recommendation Algorithms Adapting to Context Changes.The arithmetic introduces the concept of user context utility and gives an effective method to calculate the context utility difference.The difference is used for scenario pre-filtering.That reduces the use of unreasonable data and achieves the purpose of adapting to changing contexts.Experiment show that Points of Interest Recommendation Algorithms Adapting to Context Changes has significantly improved the situation-aware recommendation system.