•   When: Thursday, January 28, 2021 from 11:00 AM to 12:00 PM
•   Speakers: Preetha Chatterjee, PhD candidate, Department of Computer and Information Sciences, University of Delaware
•   Location: ZOOM
•   Export to iCal

Abstract: Software developers are increasingly having conversations about software development via online chat services (e.g., Slack, IRC, Gitter). Many of those chat communications contain valuable information, such as descriptions of code snippets and APIs, opinions on good programming practices, and causes of common errors/exceptions. The availability of this information in chats may lead to new mining opportunities for building and improving software maintenance tools. However, assessing the quality of the information in the mining source is essential for building effective data-driven software tools. Currently, there is a lack of analyses of quality assessment in developer chat platforms. Additionally, information on chat forums is shared in an unstructured, informal, and asynchronous manner, where multiple questions are discussed and answered in parallel by different participants in the same channel. In this talk, I will discuss my research focused on addressing these fundamental mining challenges, which are unique to chats.

First, I will present a supervised machine learning-based approach to identify post hoc quality conversations, i.e., conversations containing useful information for mining or reading after the conversation has ended. Automatically identifying post hoc quality conversations takes a first step in the research of quality assessment with developer chat communities. The next part of my talk will focus on techniques for mining question-answer (Q&A)  pairs from chat conversations. I will present a technique, ChatEO, that automatically identifies opinion-asking questions from chats using a pattern-based approach, and extracts participants’ answers using a deep learning-based architecture. ChatEO provides a significant contribution to using software developers’ public chat forums for building opinion Q&A systems, a specialized instance of virtual assistants and chatbots for software engineers. 

Bio: Preetha Chatterjee is a PhD candidate in the Department of Computer and Information Sciences at University of Delaware. Her research interests are primarily in software engineering, with an emphasis on improving software engineers’ tools and environments through data mining, text analysis and machine learning. She is especially interested in mining software repositories at a large scale, extending data analytics solutions to transform the plethora of information available in software artifacts into actionable nuggets of knowledge and tools, useful for both software engineers and researchers. Her research takes a significant step to positively impact new research directions in mining previously unexplored resources in software engineering, and has been published in top-tier venues including International Conference on Software Engineering (ICSE), International Conference on Mining Software Repositories (MSR), and Journal of Systems and Software (JSS).