Treffer: Automated Requirement Prioritisation Technique Using an Updated Adam Optimisation Algorithm

Requirement Engineering plays a crucial role in developing software and focuses on gathering requirements from stakeholders with diverse interests. The optimisation algorithm aims to select features to give meaningful information about requirements. These features can be used to train a model for prioritising requirements, which remains a challenging and complex task. The study aims to understand which optimisation algorithms consider suitable features and assign priority to the requirement. The study shows that the existing Adam Algorithm needs to be more capable of assigning accurate priority to the requirement due to the sparse matrix generated for the text dataset and being computationally costly. It was also unable to consider the requirements dependency. This paper suggests an improved approach to prioritise requirements called the Automated Requirement Prioritisation Technique (ARPT) to overcome the limitations of the Adam Algorithm. Compared to Adam Algorithm, the ARPT method results show a wide gap of mean squared error of 1.29 against 6.36. The execution time of the proposed method is 1.99ms as against the Adam algorithm, which is 3380ms. Based on our work described in the paper, it can be concluded that the results of ARPT in assigning priorities to requirements have improved by 80% compared to the Adam algorithm.