Research On Collaborative Filtering Recommendation Algorithm Based On User Rating And Interest

With the rapid development of Internet technology,cloud computing and smart phones,the amount of data has shown an exponential growth,and the era of big data has arrived.However,information overload has become a major problem in the era of big data,making it difficult for users to effectively acquire valuable information.In order to solve this issue,recommended system is appeared.The core idea of the algorithm is to calculate the similarity among users.The user sets whose rating is similar to the target user are found in accordance with massive user data and their rating data.Besides,these users are used as neighbors and recommendations are made through and ordered list which is organized by the users' favorite items.Nevertheless,in practical applications,the traditional collaborative filtering algorithm only uses the user's rating table to calculate the similarity of ratings among users,ignoring practical issues such as user rating differences and user interest,which leads to low recommendation quality.For the sake of solving the above-mentioned problems,this thesis has conducted research on user rating differences and user interests.The main research contents are shown as follows:1.Aiming at the problem of low accuracy of traditional recommendation algorithm,a collaborative filtering recommendation algorithm based on user rating difference is proposed.The difference of various user rating items is considered,the difference of various user rating items is taken into account as well.Firstly,the user rating difference factor is added to the traditional similarity algorithm to calculate the user rating difference.The optimum rating difference factor can be obtained through experiment.Then,the hot items punishment factor is added,which reduces the possibility of similarity caused by the hot item selection by the user,and an integrated algorithm model for the user's rating difference is obtains.Experimental results show that the algorithm improves the recommendation accuracy compared with the traditional algorithm.2.Aiming at the problem of high sparsity in collaborative filtering recommendation algorithm,a collaborative filtering recommendation algorithm based on user interest is proposed.The user's interest in the item type and the user's attributes is considers in algorithm.Firstly,interestingness of each user for each type of item and the similarity of the user's interests are calculated.Then,the user attribute model is established to calculate the similarity among user attributes.Then,these two similarities are integrated into the traditional similarity algorithm.The Movielens datasets are employed and experiments verify that the proposed improved algorithm not only enhances the accuracy of the recommendation,but also yields better results in the case of sparse data.