Analysis of Software Requirements Using Natural Language Processing

Successful development of any software system requires a set of complete, consistent and clear non-redundant requirements. Many studies demonstrate that inability to understand and manage requirements in general, as well as requirement conflicts in particular, is one of the most common causes of project cost and schedule overruns, which leads to project failure. The electronic control units in vehicles are becoming more and more complex. This is also associated with an increase in the requirements for a control unit. This large number of requirements means that some requirements are redundant or even contradict each other. There are also a number of requirements that are not fully described. Once all requirements gets cleaned then applied K-Means clustering technique to find the targets and then using supervised machine learning algorithms to classify the requirements.

Keywords: Natural Language Processing (NLP), Unsupervised Machine Learning, Supervised Machine Learning, K-Means Clustering.