The Research And Application Of Resource Recommendation Algorithm Based On Friend Relationship Analysis

Along with the advent of the Web2.0 era and the popularity of the internet, social networks has also become a platform for people to convey information and express emotions, and it has also become a way of disseminating information. However, due to a large variety of online resources, more obstacles have been in the way of helping people obtain beneficial resources. Hence, how to recommend user-friendly information has become the focus and direction of the research. In view of this problem, an increasing number of researchers yield their own distinct perspective. On the basis of the predecessors' viewpoints,this paper proposes a new recommendation algorithm to improve the recommendation effects, aiming at achieving the satisfaction of the target users.First and foremost, two important data (The users' friendship as well as history records) are obtained based on the characteristics of the social networking. What's more,the user-similarity S, one of the friends' relationship model, is introduced based on the analysis of collaborative filtering algorithm. The user-trusting T, which is put forward based on users' different preference for different online friends. The user-reputation R is proposed basing on the different importance which is caused by the users' different location in the social network. Taking that different friends' relationship models are all able to yield different influences from distinct levels into consideration, according to the different effects which are caused by the different proportion of the three friends'relationship models, the optimal proportion of the three models are selected as the users'final recommend trust,followed by determining the users' most trust collection of friends.In addition, different user's rating on a special matter based on the recommendation trust and the trust collection of friends is used to predict the target user's rating on it, hence yielding the corresponding result. Last but not least, this kind of recommendation algorithm, regarded as the core technology, helps create a recommendation system of micro learning resource?In addition to this,this paper extracts the accuracy rate and the recall rate as the evaluation index, showing that this recommendation algorithm is equipped with a relatively high accuracy rate.The recommendation algorithm which is introduced in this paper are not only adapted to the recommendation system of micro learning resource, but also applied to online educational resources' recommendation, goods recommendation in e-commerce,as well as other recommendation system which can obtain users'friendship and scoring records. Therefore, this recommendation algorithm is furnished with a wider range of applications.