Requirements Copilot: Ambiguity Management in Feature Requests

The popularity and usage of software applications (apps) among users have encouraged the industry to grow throughout many domains and categories. In today's competitive market, the requirements for apps change rapidly and continuously. Further, these requirements can be introduced through constant feature requests by end-users & other stakeholders of apps through various sources. Such requests are expressed in Natural Language (NL), where they can introduce various types of ambiguity which can lead to misunderstandings between stakeholders, developers, and testers about what feature is actually requested. This can result in wasted time and effort as developers may implement features incorrectly or build the wrong functionality altogether. This proposal aims to tackle ambiguity in feature requests using large language models (LLMs). Ultimately, the objective is to utilize LLMs to develop a practical and interactive framework to support the requirements evaluation phase. Through this framework, the submitted feature & change requests are first evaluated to detect potential ambiguities. Then, interactive negotiations with users are initiated to address such deficiencies.