Understanding and Engineering Social Dynamics
2018-12-12T21:20:17Z (GMT) by
All activities on social media evolve with time. Consequently, being able to understand<br>and engineer social dynamics, the way how various properties of social media<br>evolve, is a central question for social networks research. While recent work has studied<br>social dynamics from various angles, two important properties of social dynamics<br>are yet to be addressed, i.e., heterogeneous features and signatures at multiple time<br>scales. However, considering heterogeneous features is necessary to build a general<br>tool with wide applicability, whereas considering multiple time scales is indispensable<br>to study how social dynamics in dierent time scales interact with one another.<br>In this thesis, we aim at addressing these two properties using computational<br>algorithms with statistical groundings. In particular, we propose scalable and eective<br>methods for three basic tasks: pattern mining, structure decomposition, and datadriven<br>dynamics engineering. For each task, the proposed methods are analyzed<br>formally and veried empirically. The results reveal several interesting insights and<br>demonstrate various practical applications, such as dynamics prediction, anomaly<br>detection, and targeted intervention. Finally, the methods we propose in this thesis<br>are general enough to handle multi-dimensional time series; we have explored this<br>direction by considering other applications, such as human behavior recognition and<br>macroeconomics.