Abstract
Discovering communities in a network is a fundamental and important problem to complex networks. Find the most influential actors among its peers is a major task. If on one side, studies on community detection ignore the influence of actors and communities, on the other hand, ignoring the hierarchy and community structure of the network neglect the actor or community influence. We bridge this gap by combining a dynamic community detection method with a dynamic centrality measure. The proposed enhanced dynamic hierarchical community detection method computes centrality for nodes and aggregated communities and selects each community representative leader using the ranked centrality of every node belonging to the community. This method is then able to unveil, track, and measure the importance of main actors, network intra and inter-community structural hierarchies based on a centrality measure. The empirical analysis performed, using two temporal networks shown that the method is able to find and tracking community leaders in evolving networks. © 2020, Springer Nature Switzerland AG.

Abstract
Data stream mining seeks to extract useful information from quickly-arriving, infinitely-sized and evolving data streams. Although these challenges have been addressed throughout the literature, none of them can be considered "solved." We contribute to closing this gap for the task of data stream clustering by proposing two modifications to the well-known ClusTree data stream clustering algorithm: pruning unused branches and detecting concept drift. Our experimental results show the difficulty in tackling these aspects of data stream mining and the sensitivity of stream mining algorithms to parameter values. We conclude that further research is required to better equip stream learners for the data stream clustering task.

Abstract

Abstract
Existing approaches for detecting anomalous events in time-evolving networks usually focus on detecting events involving the majority of the nodes, which affect the overall structure of the network. Since events involving just a small subset of nodes usually do not affect the overall structure of the network, they are more difficult to spot. In this context, tensor decomposition based methods usually beat other techniques in detecting global events, but fail at spotting localized event patterns. We tackle this problem by replacing the batch decomposition with a sliding window decomposition, which is further mined in an unsupervised way using statistical tools. Via experimental results in one synthetic and four real-world networks, we show the potential of the proposed method in the detection and specification of local events. © Springer Nature Switzerland AG 2019.

Abstract

Abstract