Research On Collaborative Filtering Recommendation Algorithm Based On Item Tags And Ratings

With the rapid development of the Internet,the information that user can obtain from the Internet has grown geometrically.The Internet has changed people's lives,brought a lot of convenience,and also caused some negative effects."Information overload" is one of them."Information overload" means that it is difficult for us to quickly and accurately find the information we need from vast amounts of information.There are two technical means to solve the problem of information overload.The first is information search technology represented by search engines,and the other is information filtering technology represented by recommendation systems.The difference between the two means is that the search engine requires the user to have a clear demand.The quality of the information obtained depends largely on the accuracy of the user's description of the information.The recommendation system is different,it is based on the user's historical behavior and data to tap the user's interests and needs.It can filter out information from the mass of users interested in information.When the user's own needs are not clear,the role of the recommendation system becomes more important and can better meet the individual needs of the user.The core of the recommendation system is the recommendation algorithm.At present,many recommendation algorithms have been proposed.Collaborative filtering is the most popular and effective method among many recommendation algorithms.Although collaborative filtering recommendation algorithm has been widely used in practical commercial recommendation,there are still problems such as data sparsity,cold start and so on.This paper proposes a collaborative filtering recommendation algorithm that combines project tags.The core of the algorithm is to improve the project similarity measurement method,improve the accuracy of the recommendation,make the recommended items more personalized,and alleviate the problem of data sparseness.The best situation.The main tasks include:For the traditional project-based collaborative filtering recommendation algorithm,there is a problem of low recommendation accuracy and data sparsity.A new project similarity measurement method TPSSI was proposed.On the one hand,the tag information is introduced,the similarity between items is calculated through the tag information of the project,and the shortcomings of calculating the similarity using only the item rating information are overcome.On the other hand,the similarity of the structure and similarity of the scores are fully considered.For the similarity measurement method of the original project,when the user-item scoring information is used to calculate the similarity of the project,due to the data sparsity problem,the calculated similarity of the project is not exact.The PSSI similarity measure method was used instead to calculate the similarity between items,making the recommendation more accurate.This paper uses Movie Lens public data set to verify the experiment,and compares the improved algorithm proposed in this paper with traditional collaborative filtering algorithm based on cosine correlation and Pearson correlation-based collaborative filtering algorithm.The experimental results show that the proposed algorithm The algorithm can effectively improve the accuracy of the recommendation,and can also alleviate the poor recommendation due to data sparseness.