A Chatbot for Goal-Oriented Requirements Modeling

Abstract

Goal-Oriented Requirements Engineering approaches, in which the KAOS framework plays a key role, have been widely used for eliciting software requirements because they provide an easier way of communicating among stakeholders. However, the goal-oriented requirements modeling is not an easy way for novice requirements engineers. These professionals need more support in creating KAOS models. Recent studies have focused on the applicability of Artificial Intelligence techniques (e.g., Natural Language Processing – NLP) to support Requirements Engineering activities. In this sense, this paper aims to describe a way to support requirements elicitation for novice requirements engineers through the use of NLP within a chatbot. The chatbot (KAOSbot) acts as a KAOS modeling assistant. To evaluate our hypotheses about perceived efficacy from the novice requirements engineers perspective, we performed a quasi-experiment concerning KAOSbot’s perceived ease of use, perceived usefulness and intention to use. The results show that KAOSbot tool is a promising approach for specifying KAOS models because it was perceived as easy to use, useful, and the participants intend to use it in the future.