Improving User-experience personalisation for Everware-reproducible Science Platform

Nowadays science clearly requires a higher level of reproducibility. This can be seen from the number of re-research cases, during which people spend considerable amount of time on the repetition of the results obtained by others. The idea of reproducibility is especially important when it comes to the rapidly developing field of science. An example of such an area is machine learning. Machine learning specialists (or, more broadly, data analysis) use a significant number of tools (such as software libraries, utilities, etc.), and to obtain the same results, the researcher must use the same set of tools as the author of the work that needs to be reproduced. In order to make the research completely reproducible, we need to take care of several aspects: a description of the research protocol, access to data, preservation of the environment, a workflow pipeline and preservation of the analysis scenario. There are several platforms but neither of them on their own solves all the challenges. In addition to this, all of them have lack of personalization and user- experience. In this project, the Everware platform is overviewed, which combines the version control system (GttHub, GitLab), the virtualization technique (Docker) and the Jupyter research platform, making it possible to conduct easily reproducible studies in a collaborative manner. Result of the project is modules for Everware to work with private repositories and manage research states which improve user-experience personalization.