An Approach to Cluster Scenarios According to their Similarity using Natural Language Processing

• Autores: Delle Ville, Juliana; Torres, Diego; Fernández, Alejandro; Antonelli, Leandro
• Año de publicación: 2023
• Idioma: inglés
• Tipo de recurso: documento de conferencia
• Materia: Ciencias de la Computación e Información; Scenarios; Natural language processing; Similarity
• Institución: Comisión de Investigaciones Científicas de la Provincia de Buenos Aires

Scenarios are ideal to capture knowledge in human computer interface software engineering. Requirements engineering is a fundamental part of software development. If errors appear in this stage, it will be expensive to correct them in further stages. The domain experts and the developer team belong to different worlds. This generates a gap in communication between them. Because of it, it is important to use artifacts in natural language to communicate both sides. One simpler approach to specify requirements is Scenarios. They are widely used artifacts that generally describe the dynamics (tasks, activities) to be carried out in some specific situation. Generally, scenarios promote communication and participation from both sides. This can cause some problems. One of these problems is redundancy, that occurs when two stakeholders describe the same situation in different artifacts. This paper proposes an approach to analyze a set of scenarios by grouping them according to their similarity. The similarity is calculated through a series of comparisons of the different attributes of the scenario. This paper also describes a prototype implementing this method. Finally, the paper shows the result of a preliminary evaluation with results about the applicability of the approach.