Topic based time-sensitive influence maximization in online social networks

Abstract

With the exponential expansion of online social networks (OSNs), an extensive research on information diffusion in OSNs has been emerged in recent years. One of the key research is influence maximization (IM). IM is a problem to find a seed set with k nodes, so as to maximize the range of information propagation in OSNs. Most research adopts general greedy, degree discount, and centrality measures, etc. to find the seed node set. However, the time sensitivity is ignored in terms of propagation delay and duration under some specific situations such as emergent epidemiological bulletins and natural disaster warning. In the meantime, prior studies have mainly based on static network topology, but the user’s online/offline status and the topic preference of interaction make it difficult for traditional methods to be applied well in reality. Considering the above analysis, in this paper, we focus on the problem of time-sensitive influence maximization and give its formal description. Here, a dynamic network is constructed in terms of users’ online status. To solve the problem, we propose a topic based time-sensitive dynamic propagation model considering user topic preference and interaction delay. Besides, two algorithms are used in our work to find a seed set that could maximize the influence in OSNs with the time constraint, namely topic-based time-sensitive greedy algorithm (TTG) and topic-based time-sensitive heuristic algorithm (TTH), respectively. Extensive experiments on Twitter dataset demonstrate the efficiency and influence performance of the proposed algorithms on evaluation metrics.