Siddata Study Assistant


The interdisciplinary project “SIDDATA” examines how students can be supported in achieving individual study goals. To this end, previously unlinked data and information are combined in a digital study assistant and prepared for self-responsible use. Students can use the assistant flexibly and determine individually which factors and data sources should be considered. The data that can be used includes data from learning management systems, offers and resources of other universities and institutions, and data on individual learning and working behavior.
With the use of the assistant, students should be encouraged to define their own study goals and to follow them consistently. In the future, the data-driven environment will be able to give hints, reminders and recommendations appropriate to the situation, as well as regarding local and remote courses and Open Educational Resources (OER). These tips and recommendations should help students to make informed decisions for their own individual study path. [1]https://www.siddata.de/en/

At the universities Osnabrück, Hannover and Bremen, the employed content and study management system is Stud.IP. Through this portal, students can manage their studies, registering for courses and perform many other administrative tasks. In the course of the Joint project for Individualization of Studies through Digital, Data-Driven Assistants, a plugin for Stud.IP was developed, that leverages the vast corpus of course data as basis for a set of recommenders that help students invididualize their learning experience by finding the right courses, study partners or open educational resources.

/images/siddata_main.png

The assistant itself

My Contribution
For this project, my job consisted of implementing Machine Learning Models to suggest possibly relevant courses and other contents to students, as well as general upkeeping the backend. In the process of this, I completely re-created the project from scratch, such that it includes proper testing in a CI/CD pipeline, adheres to modern code standards, ensured both backend and ML-models scale accordingly for productive use-cases, and can be deployed easily using Docker.
Read more