Understanding and Modeling User Behavior in Social Question and Answering
Open Access
- Author:
- Liu, Zhe
- Graduate Program:
- Information Sciences and Technology
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- December 08, 2014
- Committee Members:
- Bernard James Jansen, Dissertation Advisor/Co-Advisor
Bernard James Jansen, Committee Chair/Co-Chair
John Yen, Committee Member
Heng Xu, Committee Member
James Landis Rosenberger, Committee Member - Keywords:
- social question and answering
social search
information seeking
social networks
Twitter
Weibo - Abstract:
- In this research, I focus on a specific type of information seeking on social networking sites (SNS), called social question and answering (social Q&A), in which people ask natural language questions to their networks via status updates. Compared with the traditional information seeking methods (e.g. search engines, online databases, etc.) and community question answering services (e.g. Yahoo! Answers, Quora, Answers.com, StackOverflow, etc.), social Q&A is considered to provide more trustworthy, contextualized, and interactive responses. Although there is an extensive amount of research and literature available on the topic of information seeking via the Internet, due to the distinct nature of social Q&A, still relatively little is known about how users behave in the process of information seeking and sharing under social context. To fill in this gap, in this study I examine three aspects of social Q&A. First, in order to better understand the information needs of the questioners in a microblogging environment, I develop a taxonomy of question types in social Q&A, called ASK. This taxonomy allows us to differentiate questions according to their underlying intents, and then direct them to the most appropriate respondents for help. To apply the taxonomy to practical problems, I also present the implementation of an automatic classification method to categorize the social Q&A questions according to the ASK taxonomy. Second, noticing the low response rate in social Q&A, I examine a set of factors that would affect the response probability of a question. Specifically, I limit the scope to extrinsic factors only, given the existence of literature focusing on factors from both social and cognitive perspectives in knowledge sharing. For the analysis, I collect over 20 thousand real-world questions posted on Weibo, which is the largest microblogging site in China. Third, to validate the effectiveness of question routing systems in social Q&A, as well as to characterize the voluntary knowledge sharing behavior among individuals under social context, I analyze a collection of questions and answers posted during a 10-month period on Weiwen, a Chinese question routing services based on microblogging sites. I explore the patterns demonstrated by the knowledge contributors from three different perspectives: user behavior, user interest, and user connectedness. I also propose a predictive model on active contributors in social Q&A environment based on a number of non-Q&A features. I believe this thesis is of interest as it very well addresses the research gaps concerning the understanding of information needs in social Q&A, and the effectiveness of SNS in handling question-answering. Findings from this research would be of practical value as well, as it could fulfill the need for technologies capable of performing question routing tasks to help people find what they need.