Friendships are dynamic. In this project, we uncover the dynamics of tie strength and online social interactions in terms of various aspects, such as reciprocity, temporality, and contextually. Based upon these dynamics, we build a learning to rank framework to predict social interactions in online social networks.
Dynamics of tie intimacy
- Reciprocity—Whether the tie strength is reciprocal.
- Temporality—How the tie strength evolve over an extended period
- Contextuality—How the tie strength changes based upon contexts such as a user’s current location, interest, sentiment, degree of self-disclosure, and responsiveness.
Learning to rank friends
A learning to rank framework for predicting repliers and retweeters given a particular tweet posted at a certain time.
人数
Chin-Yew Lin
Senior Principal Research Manager