In the digital age, the rapid spread of misinformation and rumors poses a critical challenge for social media platforms and users. Existing rumor containment methods often overlook the diverse range of topics associated with information and fail to consider user interests, resulting in incomplete understanding of rumor propagation. To address this issue, we introduce the Topic-aware Rumor-Truth Cascade (TRTC) model, which incorporates user interests and topic relevance to better capture the dynamics of information propagation. We define the Topic-aware Rumor Containment Maximization (TRCM) problem within TRTC model and prove its monotonicity and submodularity properties. To solve this problem, we propose Topic-aware Reverse Reachable Count (TopicRRC), an efficient index-based algorithm that leverages reverse sampling techniques to quickly identify effective truth seed sets for multiple online TRCM queries, thereby reducing both computational time and memory usage. The extensive experiments on real-world datasets demonstrate that TopicRRC outperforms existing approaches in terms of rumor containment effectiveness and computational efficiency.

With the widespread use of GPS-enabled smart devices, online social networks are increasingly integrated with offline local services. People are also more inclined to share their real-time offline location and time information with friends on online platforms. Influence maximization, which involves selecting a set of seed nodes to maximize influence within an online social network, has gained significant attention, especially on location-based social networks. However, most recent studies have focused on users’ discrete check-in data and used bipartite graphs to model the one-way relationship between online users and offline locations, without considering the interactions between different offline location. To address this gap, we introduce the location promoting influence maximization (LPIM), which aims to maximize the number of online users visiting promotional offline location. We also propose the TrajectoryCompetition algorithm, which takes into account users’ movement trajectories to capture their mobility patterns and characteristics, along with the competitive relationships between similar offline locations. Furthermore, the algorithm explores potential connections between online users to better estimate their influence. Extensive experiments conducted on datasets from six real-world cities validate the efficiency and effectiveness of the TrajectoryCompetition algorithm.

With the rapid development of science and technology, influence maximization (IM) problem has been a hot research issue. There are positive and negative relations in social networks, so the problem of IM in signed networks has a wide range of applications. Moreover, the dissemination of information is usually time-sensitive in social networks. Therefore, in this article, we propose a problem about maximizing the positive influence in signed networks within a limited time (PIMST), further we utilize the influence path to calculate the influence probability and propose an algorithm that is based on forwarding index and inverted index. In the proposed algorithm, we first select candidate seed nodes by combining two effective heuristic methods. Then, we design an algorithm to get the activation probability between node pairs in social networks. At last, we devise a method which combining forward index, inverted index with the idea of cost delay, this method speed up the selection process of seed nodes. Finally, the experimental results on three social network datasets illustrate the effectiveness of our approach compared with other algorithms.