Test case Recommendation for regression with Named Entity Recognition for test step prediction

Regression testing plays an important role in controlling the quality of the software. It, in general, accounts for a large proportion of software development costs. Therefore, if it is possible to establish any underlying relations in the existing test data, through automatic test case recommendation of relevant use cases for the regression phase, then a significant decrease in cost is feasible. In this paper, we look at existing test scripts used by our automation team. Test cases are recommended based on the priority and functionality. The priority is obtained by analysing the complexity of the test cases, and the functionality is obtained by clustering of similar test cases. Natural language processing is used to add context to step descriptions, so that, newer test cases can be designed from the context of previous similar test cases.