Diffusion-based recommendation with trust relations on tripartite graphs

Abstract

Abstract. The diffusion-based recommendation approach is a vital branch in recommender systems, which successfully applies physical processes to make recommendations for users on bipartite or tripartite graphs. Trust links indicate users' social relations and can provide the benefit of reducing data sparsity. However, traditional diffusion-based algorithms only consider rating links when making recommendations. In this paper, the complementarity of users' implicit and explicit trust is exploited, and a novel resource-allocation strategy is proposed, which integrates these two kinds of trust relations on tripartite graphs. Through empirical studies on three benchmark datasets, our proposed method obtains better performance than most of the benchmark algorithms in terms of accuracy, diversity and novelty. According to the experimental results, our method is an efficient and reasonable way to integrate additional features into the diffusion-based recommendation approach.

A Survey on Trust Prediction in Online Social Networks

“…For instance, 'the density of a typical trust network in social media is less than 0.01' [13] [33]. As another example, 'the sparsity of Advogato, Ciao, and Epinions, FriendFeed, and Flixster [frequently used datasets in trust prediction related research], i.e., the ratio of the observed trust relations to all the possible relations, is 0.0011%, 0.0028%, 0.0042%, 0.0041% and 0.0035%, respectively [34] [35] [13] [4]. It is challenging to predict the trust relations well with so limited observed links' [32].…”

TruGRC: Trust-Aware Group Recommendation with Virtual Coordinators

“…Trust is an important feature on social networks as it indicates the relationships between users. In practice, asymmetric trust is more general than symmetric trust [43]. For example, one user following another user on Twitter can be seen as a trust link between these two users, but an asymmetric one, because the trust is not mutual.…”

Personalized Recommendation via Trust-Based Diffusion

“…Novelty [31] measures the capacity of recommender systems to recommend unpopular objects. Novelty is defined by the average degree of recommended objects, written as…”

“…Deng et al [30] proposed a recommendation algorithm by combining two-step turn-around diffusion on the user-object bipartite graph and on the user-user bipartite graph, respectively. Wang et al [31] studied diffusion-based recommendation with trust relations and presented a new distribution process by combining CosRA-index [29] and both implicit and explicit trust relations. However, studies on trust-based diffusion methods on tripartite graphs remain insufficiently.…”