Skip to content Skip to navigation

PhD Defense: Extracting Users in Community Question-Answering in Particular Contexts


Community Question-Answering (CQA) services, such as Yahoo! Answers, Stack Overflow and Brainly, have become important sources of seeking and sharing information. Online users use CQA to look for information and share knowledge on topics ranging from arts to travel. The questions posted on CQA sites often rely on the wisdom of the crowd or the idea that the best answer could come from a culmination of several answers by different people with varying expertise and opinions. Given that CQA is a user-driven service, user experience becomes an important aspect, affecting the activeness and even the survival of the site. In this work, we are interested in studying the behavior of the users who participate in CQA. Specifically, we wish to understand how different types of users could be identified based on their behaviors on a CQA-specific problem at hand. A user's behavior depends on their particular contexts. For example, when we say that Alice is a “good user”, the interpretation of her behavior actually rests on the context in which it occurs. She might be a good user in the whole community, a good user for a specific topic, a good user for a particular question or a good user for a particular answer. In this dissertation, we will study and extract users in different levels of granularity. Examples of such classes of users include (i) potential answerers (ii) good/bad answerers, (iii) struggling users, and (iv) rising star users. Users are the main driving force in CQA and understanding them allows us to know the current state of their respective sites. The findings in this dissertation will be useful in identifying specific CQA user types.

Long Le
CoRE A (301)
Event Date: 
04/25/2017 - 12:00pm
Prof. Chirag Shah (Chair), Prof. Amélie Marian, Prof. Thu Nguyen, Prof. Panos Ipeirotis (NYU).
Event Type: 
PhD Defense
Dept. of Computer Science