Research On Recommendation Model And Its Application Based On User Interest Mining

With the popularization of the Internet and 5G mobile phone terminals,people leave information on the Internet anytime and anywhere,leading to rapid growth of data on the Internet.Users are submerged in the torrent of information and cannot quickly select information that is beneficial to them.This is the "information overload" problem.The recommender system provides high-quality recommendations to help users make effective choice decisions from many choices and present it to them.This is currently an effective way to solve "information overload".Since the collaborative filtering algorithm was proposed,it has become a hot research direction and has been applied in all aspects of life,but it is limited by data sparsity and cold start problems,which leads to poor recommendation effects.In recent years,with the development of mobile devices and social networks,it has become easier to obtain contextual information and social relationships.Therefore,in the user-based collaborative filtering model,this thesis considers contextual information and social relationships,and designs corresponding algorithms to solve the problem of data sparsity and cold start.The main innovations of this thesis are as follows,1)Since contextual information significantly affecting users' decisions,it has attracted widespread attention.User typicality indicates the preference of user for different item types,which could reflect the preference of user at a higher abstraction level than the items rated by user,and can alleviate data sparsity.But it does not consider the impact of contextual information on user typicality.This thesis proposes a novel context-based user typicality collaborative filtering recommendation algorithm(named CBUT),which combines contextual information with user typicality to alleviate the data sparsity of context-aware collaborative filtering,and extracts,measures and integrates contextual information.First,the items are clustered and classified into different item types.For different users,the significance of contextual information for different item types is defined and measured via knowledge granulation.Then,the contextual information is combined with user typicality to measure the context-based user typicality;subsequently,the `neighbor' users are determined.Finally,the unknown ratings under a single context are predicted,and the unknown ratings under multi-context are predicted according to the weighted summation of the significance of contextual information.The experimental results demonstrate that CBUT can effectively improve the accuracy of recommendation and increase coverage.2)The trust relationship helps to improve the accuracy of the recommendation algorithm and user satisfaction.This thesis proposes a multi-stage filling strategy algorithm under the framework of user-based collaborative filtering algorithm.Before calculating the user similarity,the user-item rating matrix is pre-filled to achieve alleviation of data sparsity and cold start problem.According to the user's original trust relationship,a degree of trust between the target user and other users is inferred under the random walk model,and then the other users are divided into three regions according to the trust relationship: direct trust region,indirect trust region and un-trust region.In the first stage,the missing values are filled in based on the ratings of users in the direct trust region;in the second stage,user typicality is introduced,and the missing values are filled based on the combination of the ratings of typical neighbors and the users in indirect trust region;in the third stage,the last remain missing values are filled in with the average user rating.Finally,the experimental results illustrate that the algorithm proposed in this thesis can effectively alleviate the problems of data sparsity and cold start.