Results of the University of Helsinki OpenReq Hackathon
3.4.2019
Results of the University of Helsinki OpenReq Hackathon
The hackathon is over. What happens next?
The Empirical Software Engineering research group at the University of Helsinki organised an OpenReq hackathon together with the Qt Company in March in Helsinki. The hackathon focused on requirements.
The participants of the hackathon were students of the University of Helsinki. As all the participants studied data science, they, naturally, emphasised data rather than infrastructure or services per se.
The first group of students completed a system that goes through the comments of issues in Jira (given in OpenReq JSON format) and proposes dependencies between issues.
The second group looked at all the text content of the issues in Jira (description, name and comments). They applied the idea to turn the text fields into vectors using GloVe vectors that made it possible to query and find similar issues, which can be duplicated issues or otherwise related.
The third group of students built a deep learning based text classification system for dependency type prediction. They trained a text classifier and a language model based on PYSIDE requirement threads to go through strings of text containing titles, descriptions and comments.
The OpenReq project is going to take advantage of the results of the hackathon. The solution of the first group is actually already being deployed by OpenReq and the one proposed by the second group will probably be deployed soon. The third solution will be tested and possibly further developed later on.