In social websites, users acquire information from adjacent neighbors as well as distant users by seeking along hyperlinks, and therefore, information diffusions, also seen as processes of “user infection”, show both cascading and jumping routes in social networks. Currently, existing analysis suffers from the difficulty in distinguishing between the impacts of information seeking behaviors, i.e. random walks, and other factors leading to user infections. To this end, we present a mechanism to recognize and measure influences of random walks on information diffusions. Firstly, we propose the concept of information propagation structure (IPS), which is also a directed acyclic graph, to represent frequent information diffusion routes in social networks. In IPS, we represent “jumping routes” as virtual arcs and regard them as the traces of random walks. Secondly, we design a frequent IPS mining algorithm (FIPS). By considering descendant node infections as a consequence of ancestor node infections in IPS, we can use a Bayesian network to model each IPS, and learn parameters based on the records of information diffusions passing through the IPS. Finally, we present a quantitative description method of random walks influence, the method is based on Bayesian probabilistic inferring in IPS, which is used to determine the ancestors, whose infection causes the infection of target users. We also employ betweenness centralities of arcs to evaluate contributions of random walks to certain infections. Experiments are carried out with real datasets and simulations. The results show random walks are influential in early and steady phases of information diffusions. They help diffusions pass through some topology limitations in social networks.

Nowadays recommender systems (RS) keep drawing attention from academia, and collaborative filtering (CF) is the most successful technique for building RS. To overcome the inherent limitation, which is referred to as data sparsity in CF, various solutions are proposed to incorporate additional social information into recommendation processes, such as trust networks. However, existing methods suffer from multi-source data integration (i.e., fusion of social information and ratings), which is the basis for similarity calculation of user preferences. To this end, we propose a social collaborative filtering method based on novel trust metrics. Firstly, we use Graph Convolutional Networks (GCNs) to learn the associations between social information and user ratings while considering the underlying social network structures. Secondly, we measure the direct-trust values between neighbors by representing multi-source data as user ratings on popular items, and then calculate the indirect-trust values based on trust propagations. Thirdly, we employ all trust values to create a social regularization in user-item rating matrix factorization in order to avoid overfittings. The experiments on real datasets show that our approach outperforms the other state-of-the-art methods on usage of multi-source data to alleviate data sparsity.