Research On Mobile Application Information Recommendation Algorithm Based On User Similarity And Topic Similarity

With the rapid development of Internet technology,users can obtain more diversified information through the Internet.However,overloading of information leads to the fact that it is difficult for users to obtain valuable information from vast amounts of information.Traditional recommendation algorithms such as collaborative filtering recommendation,content-based recommendation,recommendation based on demographic recommendation and so on have achieved good results in the internet recommendation field,This paper provides an algorithm support for mobile APP personalized recommendation algorithm based on user similarity and theme similarity.First of all,this paper analyzes the research status of the recommended technology,the problems it faces,the basic structure and the implementation strategy of the recommended technology.The principle,steps of the personalized recommendation technology and the requirements for the personalized recommendation of the mobile environment are emphatically introduced.Summarizes the recommended strategies and ideas of traditional recommendation methods.Then,based on the low accuracy and low recommendation degree of some recommended methods in personalized recommendation method,an information recommendation algorithm based on user similarity and topic similarity is proposed.The offline recommendation algorithm based on user similarity can be calculated by improving the user name attribute value and the numerical attribute value in the traditional algorithm,at the same time,the interaction behavior factors that affect the preference of the user are abstracted and merged into the user similarity calculation,and finally,the weighted Given user similarity algorithm.The online personalized recommendation algorithm based on the theme similarity firstly extracts the theme through the Latent Dirichlet Allocation from the text content preferred by the user,and then abstract the user's approval of the project,and gives the algorithm to personalize the recommendation of the user.Aiming at the sudden change of user interest,this paper proposes a personalized recommendation algorithm with complex interests,which mainly solves the problem that multiple users share the same account or the user's interest suddenly changes.According to the change of location information of users in a period of time,the user's own time and behavior preference model is constructed.Through the model,it is determined whether there is a situation in which multiple users share the same account or the interest suddenly changes.Aiming at the problem of cold start in the recommendation field,a new user's preference prediction model for information is constructed by mining the hidden interest preferences between users and information,and the preference information is predicted.Finally,the simulation shows that the proposed algorithm has high user satisfaction,and has good performance in accuracy,recall and coverage compared to the User Based Pearson Correlation Coefficient and the Item Based Pearson Correlation Coefficient,which further proves the rationality of this algorithm.