The impact of using a Domain Language for an Agile Requirement Management

• Autores: Urbieta, Matías; Antonelli, Leandro; Rossi, Gustavo Héctor; Sampaio do Prado Leite, Julio Cesar
• Año de publicación: 2020
• Idioma: inglés
• Tipo de recurso: artículo
• Materia: Ciencias de la Computación e Información; Domain Knowledge; Agile methods; User Stories; Language Extended Lexicon
• Institución: Comisión de Investigaciones Científicas de la Provincia de Buenos Aires

Context: The development of software systems is a complex activity because of its nature and the management of its construction. It is challenging to create and follow a plan. Moreover, budget overrun is a common consequence of this situation. Agile methods, like Scrum, help to mitigate this problem using incremental and iterative development. Agile methods jump start new developments, but it is difficult to be agile after several months when the software has to deal with many requirements that are scattered and tangled across several User Stories written in different Sprints. Objective: In this paper, we propose a traceability approach anchored on an index structure to access specific User Stories from a large set. Our proposed strategy has the goal to consolidate the information dispersed in different User Stories into a particular lexicon: The Language Extended Lexicon (LEL). Method: The proposed approach consists of a set of rules which extract the information dispersed in the User Stories and organize it in symbols of the Lexicon. Thus, the Lexicon supplies a consolidated and organized structure to mitigate the problem of tangled information that generates lack of traceability among different sprints. Results: We assessed how the Lexicon built by our approach improves everyday activities related to requirement management. The assessment is based on a quantitative evaluation with 36 subjects. Conclusion: The approach presents benefits for requirement tracing in agile methodologies supported by the preliminary results of the evaluation. We have developed an application (a prototype) that automates the LEL derivation rules from a set of User Stories.