Naive Bayes Classification Model for Precondition-Postcondition in Software Requirements

The quality of a test case primarily depends on the software requirements. However, manually identifying crucial elements like preconditions and post-conditions within software requirements can be time-consuming and labor-intensive. This challenge has prompted a research study to propose a novel approach for test case generation using text classification. The proposed approach involves categorizing software requirements into two labels: 'none' and 'both.' These labels indicate the presence or absence of preconditions and post-conditions in software requirements. To achieve this, the research employs the Naive Bayes algorithm, a widely used probabilistic classification algorithm in text classification tasks. The algorithm leverages two libraries, namely Scikit-learn and Natural Language Toolkit (NLTK). The Scikit-learn model proves quite effective through experimentation, achieving an impressive accuracy score of 0.86. This result demonstrates the feasibility of reducing the effort and time required for classifying test case components based on software requirements. By automating this process, the proposed approach offers a promising route for enhancing the efficiency and effectiveness of test case generation in software testing. By leveraging text classification and machine learning techniques, the proposed approach not only streamlines the identification of essential components in software requirements but also opens up possibilities for further automation and optimization of the testing process.