Research On Collaborative Filtering Recommendation Method Based On Hybrid Similarity Model

With the advent of the era of big data,it is difficult for people to find the information they need from a large amount of information.The recommendation system collects and analyzes information such as user's hobbies and habits,so that it can quickly and accurately recommend the information they need to the user.Therefore,the recommendation system is a more common method to alleviate the information overload problem.And collaborative filtering algorithm is one of the most common recommendation methods.Like other recommendation paradigms,the algorithm based on collaborative filtering also suffers from the curse of sparse data and cold-start problem.Nevertheless,we propose a Recommendation based on Collaborative Filtering and Hybrid Simi larity Model algorithm in this paper.First,the algorithm calculates the similarity of users among different items,and then describes the relationship among users,items,features and tags according to the property weights and the tag weights.Next,it a djusts the score preference between different users by setting the user preference factor and the asymmetry factor.After that,a hybrid similarity model is constructed based on the user similarity,item weights and rating preferences,and adding the user-time weight information to solve the project cold-start problem.Experiments on publicly accessible Movielens data sets demonstrate that the algorithm achieves more prominent results than other related approaches across various evaluation metrics.The specific research content is as follows:(1)For the data sparse problem of long-term retention in the collaborative filtering method of recommendation system,this paper proposes a user-mixed similarity model based on score to calculate the similarity between dif ferent users in different projects,and adjusts the rating relationship of user similarit y between different projects as weights.At the same time,in order to balance the preference between different users and improve the reliability of the model,improve the accuracy of the recommendation,set the user preference factor and the asymmetry factor as weights to adjust the model.(2)Aiming at the cold start problem in the collaborative filtering method of recommendation system,this paper proposes a calculation model based on project characteristics and project tags,which transforms the relationship between users,projects,features and tags into a relational graph mode,which makes it more intuitive to describe users.The relationship between project-features and user-project-tags,the relationship weights of project characteristics and project tags are obtained,and project feature weights are combined with user time weight information to solve the cold start problem.(3)In summary,this paper finally proposes a c ollaborative filtering recommendation method based on the hybrid similarity model,combined with the score-based user mixed similarity model and the calculation model based on project characteristics and project labels to solve the data sparse and cold start problems.The algorithm in this paper runs experiments on various published classic data sets.The results show that the collaborative filtering recommendation algorithm of mixed similarity model obtains more significant results than other related metho ds in various evaluation indicators,and effectively solves the problem of data sparseness and cold start.