Evaluation of Information Diffusion Path Based on a Multi-Topic Relationship Strength Network

Hengmin Zhu,Xinyi Yang,Jing Wei,Chao Shen
DOI: https://doi.org/10.1007/s10115-022-01794-2
IF: 2.7
2022-01-01
Knowledge and Information Systems
Abstract:Paths in online social networks play different roles in information dissemination. Ranking edges and identifying the influential edges in OSNs are imperative for intervening in or guiding information diffusion. The existing studies are limited to measuring the weight of paths in networks or the impact of paths on network connectivity while ignoring measuring the role of paths in information diffusion. In this study, we solve the problem by constructing an interactive network and calculating the information diffusion capacity (IDC) of paths. First, a multi-topic relationship strength network is constructed to describe users’ interactive preferences. Second, we propose the IDC method to measure the IDC of paths considering users’ interactive activity, relationship strength, users’ influence and users’ topic preferences. Users’ multi-topic relationship strength makes it possible to distinguish the importance of edges under different topics. Finally, we compare the IDC method with the baseline methods including the connectivity-based method and the method of edge betweenness centrality from the perspectives of ranking granularity and spreading ability in two subnets from Sina Weibo and Twitter. The results show that the IDC method ranks edges in finer grain than the connectivity-based method. Based on the SIR model, we compare the spreading ability of the three methods and find that the important edges evaluated by the IDC method behave better in topic spreading. Application exploration shows that the backbone network and the weak ties in information diffusion can be discovered through evaluating the paths in the relationship strength network.
What problem does this paper attempt to address?