The Research And Implementation Of Recommendation Algorithm Based On User Characteristics And Trust

With the rapid development of the Internet,the amount of information increases explosively,which leads users to be unable to obtain really useful information in the face of a large amount of information.The appearance of recommendation system solves the problem of information overload to a great extent.Collaborative filtering recommendation algorithm is widely used in the field of recommendation because of its excellent recommendation quality.Collaborative filtering recommendation algorithm mines preferences of users by analyzing the historical behavior,finds users who share the same interests as the target users,synthesizes their evaluation,and predicts scores of the target user on the specified items.In the real environment,with the rapid increase of users and items,problems such as cold start,data sparsity and timeliness are gradually exposed,which seriously affect the recommendation quality of collaborative filtering recommendation algorithm.Aiming at the above problems,the collaborative filtering recommendation algorithm is deeply studied in this paper.The recommendation algorithm based on user characteristics and trust is proposed,which is called User-CT algorithm.The main contents of this paper are as follows:1.In view of the low accuracy of similarity calculation results caused by different user rating scales and sparse scoring matrices,the subjective scores of users are normalized;a user dynamic interest preference model is established to further mine users' real preferences from users' high scores,and time factors are taken into account to ensure the timeliness of recommendation.2.The traditional collaborative filtering recommendation algorithm measures the similarity between users by common scoring items among users,ignoring the potential relationship between users who do not have common scoring items,which also has a vital impact on the recommendation results.Therefore,a user trust relationship model is established to mine the relationship between users who do not have common scoring items.Fusion of user trust based on improved user similarity can make prediction score more accurate and data sparsity problem eased.3.The traditional collaborative filtering recommendation algorithm only relies on the user's historical behavior to recommend,ignoring the user's own attributes.The user's attributes are also related to the user's preferences.Therefore,adding the user's demographic attributes prediction score can alleviate the user's cold start problem to a certain extent.4.Comparing User-CT with traditional collaborative filtering recommendation algorithm on Movie Lens dataset,the experimental results show that the proposed algorithm performs better than the traditional collaborative filtering recommendation algorithm in recommendation accuracy and performance.Some suggestions on the selection of weighted parameters are given in combination with experiments.5.Taking the User-CT algorithm as the core,using the SSM framework and Bootstrap framework,we design and implement a complete food recommendation system.