Distributions-Based Approach to Predicting Response Times on Stack Overflow
Public Deposited- Last Modified
- February 26, 2019
- Creator
-
Arunapuram, Preeti
- Affiliation: College of Arts and Sciences, Department of Computer Science
- Abstract
- Stack Overflow is an online forum on which people can post and respond to questions concerning computer programming. Two major concerns arise from this domain: liveness (will a person respond to a post?) and speed (how long will it take for someone to answer?). Though much work has been done in studying factors that influence liveness and speed, no work has been done to predict response times. In this study, we propose a distributions-based approach to predicting the range of time in which a question is likely to be answered. We started by first reinvestigating features that were found to affect liveness and speed in other domains, and then applied them in prediction methods to gauge their influence. Namely, we used post title lengths and subject tags in response time prediction. When it came to predicting accepted answer response times, we found that including features in the prediction algorithm did not significantly improve results. However, the opposite is true in predicting earliest answer response times. More specifically, title lengths were found to be useful.
- Date of publication
- spring 2014
- Keyword
- DOI
- Resource type
- Rights statement
- In Copyright
- Note
- Funding: NSF grant IIS-0810861: HCC-Small: Collaborative Mixed-Initiative Access Control
- Advisor
- Dewan, Prasun
- Degree
- Bachelor of Science
- Honors level
- Highest Honors
- Degree granting institution
- University of North Carolina at Chapel Hill
- Extent
- 23
- Date uploaded
- April 26, 2014
Relations
- Parents:
- In Collection:
This work has no parents.