PriseBot - A chatbot to assist in the development of iStar

  • Erlânio Freire UFC
  • Enyo Gonçalves UFC
  • Marcos Oliveira UFC
  • Luiz Guilherme Moreira Leite UFC
  • Sabrina Ketlen Colares Santos UFC

Resumo


Context: The representation of requirements through iStar models often triggers the need to create extensions. The PRISE process supports the creation of iStar extensions. Problem: Developing new iStar extensions can take time and effort. That can be related to the constructs used, searching for existing extensions, or even associated with PRISE’s process and sub-processes flows. A tool is needed to make this process more interactive, effective, and productive. Solution: This work aims to present PRISEBot, a chatbot capable of extracting knowledge from the PRISE Process, a collection of existing extensions and offering constructors already used in proposed extensions. When dialoguing with the Bot, the user has access to information via a web interface or Telegram App, where they get satisfactory, high-quality answers close to the ones given by human beings quickly, efficiently, and interactively. SI theory: PRISEBot was developed with the idea of Design Theory in mind, using its knowledge base to evaluate the existing and proposed tools for its goal of facilitating iStar extension proposals. Methods: This research has a prescriptive aspect and was evaluated by conducting analyses between the main LLMs available (ChatGPT, Bing.AI, and Bard). Experts in the field empirically estimated the comparison between questions and answers. Results: During our evaluation, we asked 30 questions to each Bot. Bard answered one question correctly, totaling 3.3% assertiveness. ChatGPT could answer only one question effectively, obtaining 3.3% assertiveness. Bing answered four questions correctly, totaling 13.3% assertiveness. PriseBot answered all the questions effectively, totaling 100% assertiveness. Contributions and impact in the IS area: iStar extensions can be related to various domains when applying goal-oriented requirements analyses. Therefore, it is an essential area in Information Systems. This work aimed to develop a chatbot capable of answering iStar extender questions to produce new extensions efficiently. This chatbot can integrate with the PRISE tool and Telegram.

Palavras-chave: Engenharia, GORE, ISTAR, NLP, PRISE, Rasa, chatbot
Publicado
20/05/2024
FREIRE, Erlânio; GONÇALVES, Enyo; OLIVEIRA, Marcos; LEITE, Luiz Guilherme Moreira; SANTOS, Sabrina Ketlen Colares. PriseBot - A chatbot to assist in the development of iStar. In: SIMPÓSIO BRASILEIRO DE SISTEMAS DE INFORMAÇÃO (SBSI), 20. , 2024, Juiz de Fora/MG. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2024 .