Wed 17 Aug 2022 19:20 - 19:40 at Dibbler - Natural Language Processing for RE Chair(s): Tong Li

Pre-requirements specification (pre-RS) traceability focuses on tracing requirements back to their sources. Pre-RS traceability is an under-explored area of research, compared to post-RS traceability. Likely reasons for the limited studies are the scarcity of pre-RS resources, e.g., requirements elicitation conversations such as interviews or workshops, and the challenges of linking requirements to informal, unstructured text. Building on the increasing use of digital communication tools that allow the recording and transcription of conversations, we explore the opportunity of linking requirements to the transcript of a requirements elicitation conversation. We introduce Trace2Conv, a prototype tool that aims at tracing user story requirements back to the relevant speaker turns in a conversation. Trace2Conv makes use of NLP techniques to determine the relevant speaker turns. As an initial validation, we take automatically generated transcripts from real-world requirements conversations, and we assess the effectiveness of Trace2Conv in supporting the process of identifying additional context for the requirements. The validation serves as a formative evaluation that guides the evolution of Trace2Conv and as a inspiration for future research in the field of conversational RE.

Wed 17 Aug

Displayed time zone: Hobart change

19:20 - 20:10
Natural Language Processing for RERE@Next! Papers / Journal-First at Dibbler
Chair(s): Tong Li Beijing University of Technology
Back to the Roots: Linking User Stories to Requirements Elicitation Conversations
RE@Next! Papers
Tjerk Spijkman Utrecht University, Fabiano Dalpiaz Utrecht University, Sjaak Brinkkemper Utrecht University
Empirical evaluation of tools for hairy requirements engineering tasks
Dan Berry University of Waterloo