Research On Recommendation Algorithm Based On User Interest Partition

In recent years,the recommendation system as an effective means has been gradually adopted by most companies in terms of information overload,the core of the recommended system is the recommendation algorithm,and the core of the recommended algorithm is to calculate the similarity between users.In order to improve the accuracy of this similarity calculation,more and more people are concentrating on this issue.In the paper,a method of calculating film interest degree is proposed for the sparseness problem in the field of film recommendation.This method can improve the sparseness of data in the film recommendation and can dig out the user's interest in calculating the similarity of the user.In this paper,we experiment on the data set recommended by the film,and the experimental results show that our recommendation accuracy is improved to some extent.The user interest degree proposed in the paper can effectively dig out the potential interest of users.This method can effectively dig out the user's potential interest.First,when this algorithm is used in conjunction with clustering,it improves the calculation of user similarity,but brings the problem of time efficiency.In order to avoid this problem,this paper presents a user interest classification algorithm.In this process,we classify the users and divide the users into sub-categories.Secondly,we have improved the similarity calculation formula in order to avoid redundancy when calculating the similarity.This method can filter out some "pseudo similar" users.Through the experiment we can see that this calculation model can improve the calculation efficiency and improve the accuracy.The innovation of the paper includes the following aspects.First,the paper puts forward a method to calculate the interest degree of the users in the film recommendation area.Secondly,we classify the interest degree matrix and then use the cooperative filtering algorithm to recommend it.The above model can effectively improve the sparsity of data.Finally,in order to improve the efficiency of the calculation and in order to maintain the stability of the recommendation accuracy,we improved the similarity calculation formula that to add the threshold.The research work shows that the above model can dig more comprehensive user interests,and has certain guiding significance for the development of recommendation algorithm.