PhD Thesis Proposal --Akira Matsui

Date and Time: 1:00pm-2:30pm

Friday, Sep, 3rd

Committee: Emilio Ferrara, Nakano Aiichiro, Aram Galstyan, Xiang Ren, Marlon Twyman

Title: Mining and leveraging temporal structures of human behavior in Online Platforms

Abstract:

With the development of technology in recent years, detailed data on human temporal behavior has become available. Many methods have been proposed for mining these human dynamic behavioral data, which have provided valuable insights for research, especially in computational social science. However, most methods analyze only sequential behavior and do not comprehensively consider inter-temporal information such as time intervals between users' actions. In this proposal, we leverage the time intervals of actions to understand human behavior from a mid-term perspective. We first argue that the engagement analysis without considering time interval differences among individuals may return a biased result and propose a method to mitigate those biases. Then we also propose an embedding-based approach to incorporate inter-temporal information into human behavior analysis, which enables us to mine the temporal context of action in a unified manner. Our proposal will demonstrate the importance of inter-temporal information, especially in the application of computational social science.

Zoom: https://usc.zoom.us/j/98321628121

WebCast Link: https://usc.zoom.us/j/98321628121

Audiences: Everyone Is Invited

Contact: Lizsl De Leon