Crowd sensing harnesses the power of the crowd by mobilizing a large number of users carrying various mobile and networked devices to collect data with the intrinsic multi-modal and large-volume features. With traditional methods, it is highly challenging to analyze the vast data volume generated by crowd sensing. In the era of big data, although several individual-oriented approaches are proposed to analyze human behavior based on big data, the common features of individual activity have not been fully investigated. In this article, we design a novel community- centric framework for community activity prediction based on big data analysis. Specifically, we propose an approach to extract community activity patterns by analyzing the big data collected from both the physical world and virtual social space. The proposed approach consists of community detection based on singular value decomposition and clustering, and community activity modeling based on tensors. The proposed approach is evaluated with a case study where a real dataset collected over a 15-month period is analyzed.
Stored only when both papers are in the corpus — 154 citations exist upstream.