Meena Malu

Title:AI in testing.WHY: Idea whose time has come 

Abstract:

Great restructuring

The world we live in as it negotiates the pandemic, the business world is undergoing “Great Restructuring”. Every aspect of our life and work is going through a change. The way we interact, collaborate, learn, deliver and expect value is changed forever. It is going to have a profound impact on the future of work, workplace, and workforce. The ‘Great Restructuring’ is comprehensivebecause no industry, no company, and no country is isolated. Everyone must adapt or cede their position. Digital transformation at pace is the most critical imperative facing all the enterprises across sectors and continents, to stayrelevant post the Great restructuring.

SDLC Data Store –gold mine awaiting excavationEvery software project generatesenormous data. In terms of requirements documents, user stories, test cases, code files, user manuals, defect logs, and so on. Currentlythis data goes abegging and is barely exploited to help improve software testing and software development productivity. While the sophistication of Machine Learning and Natural Language Processing (NLP) in particular, now allow to extract rich contextual insights through building correlation and traceability across the software development life cycle.

Paradigm shift in NLP

While research and progress in the art of NLP has been incrementally improving for several decades, yet a step change has happened inthe art of NLP in the last decade. Traditional models n-gram, bag of words, topic modelling, RNN and LTSM of NLP are being upended by word embeddings and transformer models. These new models not only allow semantic but also contextual understanding of written language. Incidentally in Mar 2019, NLP deep learning achieved a new milestone of surpassing human performance on the test for the first time (86.8% accuracy) (source: GLUE leadership board).