Collaborative Filtering Recommended Algorithm Research Based On User Trust

Nowadays,the times are advancing,and science is moving from being immature to mature.The Internet has caused people's lives to undergo earth-shaking changes,and it occupies an irreplaceable place in people's daily lives.The large number of online information,and the uneven quality of the features,make it difficult for users to filter out a large amount of information that is useful to them.Faced with the annoyance brought by massive information to users,the emergence and advancement of the recommendation system has made people see hope and dawn.At present,the recommendation technology is still in a state of vigorous development.The collaborative filtering recommendation technology is still the most widely used and popular core technology in the recommendation system.It is a targeted recommendation for the user,based on a series of behaviors that the user has made in the past,and the user's own interest preference information that has been excavated through these behaviors.When the initial collaborative filtering recommendation algorithm makes recommendations for users,it mainly calculates the similarity between users and users,or between projects and projects,and the similarity is calculated based on the user's project.Previously given scoring information to dig out.This method itself has great deficiencies,which leads to a large error in the prediction results of the method,and the accuracy is low.In daily life,when we select and grade items,we often receive recommendations and suggestions from people we are familiar with and trust.This makes trust as an influential factor added to the research of recommended technologies.At present,the recommendation technology that integrates the trust factors has also been discussed and studied by many scholars,but there are also defects and deficiencies in each of them,making the recommended performance and accuracy still have room for improvement.Aiming at a series of problems existing in the traditional collaborative filtering recommendation algorithm,in this paper,the original recommendation algorithm incorporates the trust relationship between users,and a new recommendation algorithm is obtained to improve the existing inadequacies of the original algorithm.The main research content is as follows:(1)When selecting a neighbor set for a user,the conventional collaborative filtering algorithm does not consider too much of a complicated relationship between the users,but simply calculates the similarity between the users through a certain method.This will lead to the final recommendation of the result is not ideal,and sometimes there will be a particularly large error.This article gives a comprehensive consideration of whether there is a common scoring item between users.In these two cases,the calculation method of the degree of trust between two different users is given separately,and then get the trust relationship between the users synthetically.This makes the cold start and data sparsity of traditional collaborative filtering algorithms effectively solved.(2)Finally,the new recommendation algorithm presented in this paper is verified on the MovieLens dataset.The value of the parameters in the algorithm is determined by experiments.Finally,the proposed algorithm is compared with other algorithms on various indicators to prove the effectiveness of the algorithm.