Collaborative Filtering Recommendation Based On Fuzzy Clustering And Improved Shuffled Frog Leaping Algorithm

With the rapid development of the Internet,there are more and more network information,and there has been a serious information overload phenomenon.As a result,it is difficult for people to find the information they want.Therefore,recommendation system has emerged,and the recommendation system can help users quickly and unscrupulously.finding the information they need,alleviates information overload.Collaborative filtering is currently one of the most widely used personalized recommendation technologies.It has achieved considerable success in practical applications.However,because actual data is often very sparse,the problems faced by collaborative filtering recommendation algorithms are becoming more and more prominent.There are changes in user interest,data sparsity,cold start,and data expansion.The traditional collaborative filtering recommendation algorithm only considers the similarity between users(or projects)to recommend to users,but ignores the fact that user interest changes over time.For this defect,a time-based exponential forgetting function is constructed according to Ebbinghaus forgetting curve.The user score is processed by this function to obtain a weighted score matrix based on time decay.The cold start problem can be regarded as the extreme case of data sparse.For the problem of data sparseness,a fuzzy C-means algorithm is introduced to perform fuzzy clustering for users.Users can belong to different user classes with different degrees of membership,effectively increasing data.Density reduces sparsity and can solve the problem of low recommendation quality due to inaccuracy of similarity calculation.As an unsupervised machine learning,clustering can help users find similar neighbors faster and improve the algorithm's recommendation speed without losing accuracy.Shuffled frog leaping algorithm is one of the popular algorithms to solve the problem of combinatorial optimization.The algorithm has the advantages of simple thought,few parameters,fast searching speed and easy implementation,but it also has the disadvantages of slow convergence speed and easy to fall into local optimum.For its defects,the frog's simple movement rules are self-adaptively improved,and the improved algorithm is applied to the recommendation system,and the user's nearest neighbor set can be quickly found.Experiments were performed on MovieLens dataset using R language.The resultsshow that compared with the traditional collaborative filtering recommendation algorithm,the algorithm proposed in this paper effectively mitigates the adverse effects of data sparsity,and at the same time has the speed and precision of recommendation.Obviously improved.