High Dimensional Fuzzy C-Means Clustering Recommendation Algorithm Based On Density Canopy

As Internet technology continues to develop and data and information becomes more and more available,it is increasingly difficult for users to access valuable information.In order to help users find the information they need quickly,recommendation algorithms have become very important techniques that are gaining wider and wider attention.Fuzzy C-means are applied in recommendation algorithms because of its simple and flexible algorithm and efficient clustering,but the algorithm has limitations such as the need to manually predetermine the number of clusters,unstable selection of initial cluster centres,and difficulty in handling dimensional disasters caused by complex and high-dimensional data.The collaborative filtering recommendation algorithm has shown good performance in ecommerce websites,social networking sites and news pushing due to its advantages of high efficiency and accuracy,but this recommendation algorithm still suffers from problems such as sparse data,cold start,and user subjectivity affecting the recommendation results.Therefore,the following research is conducted in this paper:To address the problems that the fuzzy C-means(FCM)clustering algorithm needs to determine the number of clusters manually in advance leading to poor self-adaptability,unstable initial cluster center selection leading to low stability,and failure of similarity measure when clustering complex high-dimensional data,an FCM clustering algorithm with isometric feature mapping and density Canopy optimization(Fuzzy C-Means based on density Canopy and manifold ISOMAP,DM-FCM).Firstly,this paper proposes a density Canopy algorithm based on improved local density to automatically determine the number of clusters and initial cluster centers to improve the self-adaptability and stability of the algorithm.Secondly,considering that high-dimensional data often present a nonlinear structure,the isometric feature mapping algorithm is used to construct a stream-shaped spatial structure,which preserves the global geometric properties of complex high-dimensional data and improves the clustering effect of the algorithm on complex high-dimensional data sets.The experiments use FM index,weighted average of homogeneity and completeness,adjusted mutual information,and adjusted Rand coefficient as the performance measures of the clustering algorithm,and the results show that the isometric feature mapping algorithm is a better similarity measure,and the improved FCM algorithm improves the AMI performance measure by an average of 6.99%,the ARI performance measure by an average of 13.01% and the V-measure performance measure by an average of 6.27% compared to the traditional FCM algorithm.To address the problem that collaborative filtering recommendation algorithm can lead to accuracy degradation and low recommendation efficiency when the recommendation is performed with sparse data,we propose a collaborative filtering recommendation algorithm based on density Canopy algorithm and isometric feature mapping with fuzzy C-means(Collaborative filtering algorithm based on DM-FCM,DMFCM-CF).The density Canopy algorithm and isometric feature mapping are used to optimize the FCM clustering algorithm and improve the clustering efficiency.The FCM algorithm based on density Canopy and isometric feature mapping is used to fuzzify the users and reduce the computational effort.A matrix padding algorithm based on the neighbourhood approach is used to alleviate the data sparsity problem.The collaborative filtering recommendation algorithm based on density Canopy algorithm and isometric feature mapping makes the recommendation more accurate,effectively alleviates the problems of data sparsity and data high dimensionality,and the recommendation error is reduced with good recommendation effect.