about this project
Invicta - STARS (Software Testing Anomaly Recognition System)
It is an indisputable fact that the use of technology has become ubiquitous and prevalent in every single facet of modern life. As such, software has become the virtual backbone of human civilisation, a key component in all imaginable sectors, from healthcare and communication to commerce and entertainment. With such a significant impact software has on society, it is imperative that the process of software development is both efficient and free of defects to ensure the continual growth of industries and services. To this end, various techniques have been developed to not only optimise and standardise the software testing process, and to automate it. This allows developers to rapidly run tests on their solutions on multiple granularity levels to ascertain a certain level of functionality and stability before production. This automated testing approach, however, is yet to be perfect, as there still exists a bottleneck in the software testing pipeline, where human analysis of testing logs is still required to analyse and identify anomalous behaviours in the software. This manual log analysis process is often time-consuming and could potentially give rise to further problems.
To address these limitations and allow the development of software to be even further streamlined, our team has worked with our industry partners at Katalon to procure expert domain knowledge on software testing and identify the most vital and impactful aspects of log analysis that could potentially vastly accelerate the workflow of developers. Armed with these critical insights and the advent of Machine Learning, our team present STARS – Software Testing Anomaly Recognition System – a solution that is capable of analysing software testing logs and detecting anomalous test case occurrences through time, as well as generating suggestions for further remediation. The solution also comes with a continuous training pipeline for future alterations to the distinctive Machine Learning models for separate users.
Related Projects
