Research On Personalized Recommendation Technology Based On Fuzzy Clustering

With the continuous development of Internet technology and the exponential growth of Internet data,how to provide users with real-time and accurate personalized information services has become a hot research topic nowadays and also an urgent problem to be solved in current e-commerce websites.Therefore,personalized recommendation technology came into being,mainly based on the user's historical data to the user recommended information in line with user interests and hobbies,so as to meet the user's demand for personalized information.Collaborative filtering algorithm is one of the most studied and widely used techniques in personalized recommendation technology.This dissertation mainly focuses on user-based collaborative filtering recommendation algorithm,analyzes the existing problems in the algorithm,deeply studies on the data sparsity,and proposes a personalized recommendation algorithm based on fuzzy C-means clustering combined with fuzzy clustering.Based on the traditional fuzzy C-means clustering algorithm,the distance calculation formula in the fuzzy C-means clustering is optimized and the weighted Euclidean distance is calculated by using the Slope One algorithm to pre-fill the matrix,thereby reducing the sparseness of the scoring matrix.The formula replaces the traditional Euclidean distance formula.The specific recommendation process is to cluster the user-item scoring matrix first by using the improved fuzzy C-means clustering algorithm to reduce the matrix dimension,and then use the Slope One algorithm to fill the scoring matrix to reduce the sparsity of the matrix.Then,Collaborative filtering algorithm for the target user to complete the project recommended work.In the last part of the dissertation,the improved algorithm is simulated experimentally,comparing the recommended accuracy of each algorithm in different situations,and verifying the actual performance of the improved recommended algorithm in this paper.And using the open source and efficient recommendation engine Taste provided by Apache Mahout to build a movie recommendation system,the recommended algorithm proposed in this paper is applied to quickly and accurately recommend a movie of interest to a user according to a user's historical score.