Software Requirements Categorization Using NLP Techniques

Abstract

Categorizing software requirements is a crucial aspect of software engineering, given the rising need for software worldwide. However, manual categorization is prone to errors and can be time-consuming, leading to a need for more efficient and accurate methods. To address this challenge, a research project was conducted to identify the most effective approaches for acquiring and categorizing software requirements. The study compared several natural language processing techniques and machine learning algorithms and found that the Bag of Words method followed by the Multinomial Naive Bayes algorithm provided the best results with an F-measure of 0.93.

The findings of the study have practical implications for software engineers, enabling them to better understand the requirement categorization process and improve the accuracy of their models. Additionally, the study's proposed future work in enhancing the approach and tweaking algorithms could lead to more efficient categorization of software requirements, with significant impacts on the software industry.

In conclusion, the research project provides valuable insights into improving the process of categorizing software requirements, paving the way for faster and more accurate software development. The study's findings can inform further research in the field of natural language processing and machine learning, and the proposed future work could lead to even more efficient and accurate categorization methods in the future. Overall, the project represents a significant step towards more efficient and accurate software development processes.