Recommendation Algorithm Based on Graph-Model Considering User Background Information

With the development of information technologies and increase scale of digital resources, personalized recommendation systems have come into the big picture of web2.0 technology. This paper proposed a graph-based recommendation algorithm using the user-resource rating data to construct a graph model and improves the model by adding user background information. The Random Walk with Restarts algorithm is applied to generate the final recommendation set. The improvement in accuracy on sparse data is illustrated by the experiments on the Movie Lens data set, comparing with the collaborative filtering algorithm.