Automated Test Case Generation from User Specifications using Natural Language Processing

Abstract

Natural Language Processing is a well-known approach of Artificial intelligence which is used to extract the important information automatically from ambiguous and unstructured raw text. The approach of NLP is applied to various fields like in bio-medical domain and sentimental analysis etc. In this research thesis, we comprehensively investigate the application of natural language processing over the testing phase of software development life cycle. A Systematic Literature Review (SLR) is carried out to select 16 studies published during 2005-2016. Subsequently, 6 combinations of main NLP activities (i.e. Tokenization, POS tagging, Chunking and Parsing) are identified. Moreover, 9 NLP algorithms are found and overall 10 tools, proposed by the researchers are analyzed. 8 existing tools, utilized by the researchers in the given research context, are identified. A Novel Natural Language Processing (NLP) approach is presented in this research thesis to generate the test cases automatically from natural language software requirements. Automated Requirements to Test case Generation (AR2TG) tool is developed as a part of this research study. For the sake of evaluation, validation is performed through benchmark case studies. The experimental results prove that the proposed natural language processing approach is novel and fully automated which also improved the results of existing research studies.