Analysis

Digital traces, including online communications (e.g., emails, social media), cellphone records, collaboration data (e.g., scholarly publications), and consumer behavior data, offer tremendous opportunities to study human social behaviors. Our research incorporates multidisciplinary methodologies, including network science methods, linguistic analysis, data mining and mixed methods data analysis, to characterize the patterns of individual and collective activities at a societal scale. These studies reflect the regularities and changes in various aspects of human and social dynamics. Our work has focused on tracing collective attention and social processes in response to disaster events, as well as on making predictions about how certain disasters or disruptions could develop.

Selected publication
  • Lin, Y.-R., Margolin, D., Wen, X. (2017). Tracking and Analyzing Individual Distress Following Terrorist Attacks Using Social Media Streams. Risk Analysis (doi:10.1111/risa.12829) PDF
  • He, X., Lu, D., Margolin, D., Wang, M., Idrissi, S., Lin, Y.-R. (2017). The Signals and Noise: Actionable Information in Improvised Social Media Channels During a Disaster. In Proceedings of Web Science 2017 (WebSci 2017) (doi: 10.1145/3091478.3091501) PDFdataset
  • Chung, W.-T., Wei, K., Lin, Y.-R., Wen, X. (2016). The Dynamics of Group Risk Perception in the US After Paris Attacks. In Proc. of the 8th International Conference on Social Informatics (SocInfo 2016) (Best Paper Award) (doi: 10.1007/978-3-319-47880-7_11) PDF
Mining

We are developing new data mining approaches for temporal and structural analyses of multi-relational social networks. Our research incorporates graph mining, tensor analysis, and other statistical learning methods including deep learning. The applications include urban analytics, automatic discovery of disaster impact, anomaly detection from multi-source social systems, and link prediction in dynamic networks.

Selected publication
  • Wen, X., Lin, Y.-R., Pelechrinis, K. (2016). PairFac: Event Analytics through Discriminant Tensor Factorization. In Proc. of The 25th ACM International Conference on Information and Knowledge Management (CIKM 2016) (doi: 10.1145/2983323.2983837) PDFdataset
  • Teng, X., Lin, Y.-R., Wen, X. (2017). Anomaly Detection in Dynamic Networks Using Multi-View Time-Series Hypersphere Learning. In Proc. of The 26th ACM International Conference on Information and Knowledge Management (CIKM 2017) PDFcode
  • Li, L., Tong, H., Cao, N., Ehrlich, K., Lin, Y.-R., Buchler, N. (2017). Enhancing Team Composition in Professional Networks: Problem Definitions and Fast Solutions. IEEE Transactions on Knowledge and Data Engineering (TKDE), 29(3), 613 – 626 (doi: 10.1109/TKDE.2016.2633464)
Visual Analytics

Visual analytics aims to offer tools for people to find latent information in a complex situation, in an intuitive manner. Recent available datasets (including social media data) are usually big in volume, dynamic in time, and have rich contexts with heterogeneous data types and attribute values, which makes the traditional visualization techniques (e.g., graph/network representation) not applicable. We are developing interactive visualization techniques based on novel abstractions of heterogeneous data and the integration of human-machine interaction theories and data mining methods.

Selected publication
  • Cao, N., Lin, C., Zhu, Q., Lin, Y.-R., Teng, X., Wen, X. (2017). Voila: Visual Anomaly Detection and Monitoring with Streaming Spatiotemporal Data. IEEE Transactions on Visualization and Computer Graphics (TVCG) PDFvideo
  • Du, F., Cao, N., Lin, Y.-R., Xu, P., Tong, H. (2017). iSphere: Focus+Context Sphere Visualization for Interactive Large Graph Exploration. In Proceedings of the ACM SIGCHI Conference on Human Factors in Computing Systems (CHI 2017) PDF
  • Cao, N., Lin, Y.-R., Li, L., Tong, H. (2015). G-Miner: Interactive Visual Group Mining on Multivariate Graphs. In Proceedings of the ACM SIGCHI Conference on Human Factors in Computing Systems (CHI 2015) PDF