News

We have done a lot of groundwork on which current and future CATALPA research can build," says Prof. Dr. Jörg Haake. The Professor of Cooperative Systems in the Faculty of Mathematics and Computer Science led the LA DIVA project, which stands for Learning Analytics for Diversity-Inspired Adaptive Learning. As part of his now-completed dissertation, Ph.D. student Marc Burchard built a platform that enables experimental studies with groups within Moodle. This allows researchers to study students' learning behavior under real laboratory conditions in real courses. "It was very important that the platform was robust, so that it wouldn't break down when there were a lot of participants at the same time," he says.

Foto: Hardy Welsch

The platform has now been used for five semesters in the distance learning course "Introduction to Scientific Work in Psychology". It allows researchers to study students' activities in collaborative writing tasks: How do students collaborate and communicate? The CATALPA research project Div-Adapt, led by Prof. Dr. Stefan Stürmer, used this to investigate whether there are differences in collaboration when learning groups are composed of homogeneous or heterogeneous learners, for example in terms of socio-demographic characteristics or prior knowledge. One finding was that groups whose members had both different sociodemographic characteristics and very different levels of prior knowledge worked less closely together.

The platform is designed to assemble groups in a variety of ways, depending on the research approach and question. "Either the students themselves choose a study group, or the groups are formed randomly. The platform is designed to put together groups in a variety of ways, depending on the research approach and research question. "Either the students themselves choose a study group, or the groups are formed randomly. But it is also possible to make certain specifications and then have the groups put together by an algorithm," says Haake. For this part of the research project, the CATALPA researchers collaborated with Prof. Dr. Niels Pinkwart and his team at the Humboldt University in Berlin. "They have developed an algorithm that enables the automated composition of groups according to certain specifications." CATALPA's follow-on research projects are already using the platform's capabilities. For example, the DFG-funded MULTIDIVERSE-CSCL project is investigating the influence of stereotypes in virtual learning groups. Recently, the platform was expanded to include a video chat function with automated transcription, which in turn enables new research approaches. and then have the groups put together by an algorithm," says Haake. For this part of the research project, the CATALPA researchers collaborated with Prof. Dr. Niels Pinkwart and his team at Humboldt University in Berlin. "They have developed an algorithm that enables automated group formation according to specific requirements. CATALPA's follow-on research projects are already using the platform's capabilities: for example, the DFG-funded MULTIDIVERSE-CSCL project is using it to study the influence of stereotypes in virtual learning groups. Recently, a video chat function with automatic transcription was added to the platform, which in turn enables new research approaches.

Another part of LA DIVA dealt with automated feedback processes. "Usually we know that teachers give feedback to learners when they have completed tasks," explains Haake. "We know that feedback stimulates the learning process. But in large virtual learning groups, such as we have at the FernUni, individual feedback on exercises is hardly possible." For this reason, doctoral student Regina Kasakowskij investigated how such feedback could be automated for certain exercises - so-called self-assessments - in computer science courses. Teachers provide assessment criteria and feedback on how to meet the criteria. Depending on which of these criteria a solution meets according to the student's self-assessment, the student automatically receives the appropriate feedback. Students can then report how helpful this feedback was for them. This makes the feedback process bidirectional, with opportunities for improvement in both directions.

Follow-up projects

This part of LA DIVA has also led to follow-up projects: The Foundation for Innovation in Higher Education is funding the two Freiraum projects "METALADIN" and "Aiducator". These closely related projects aim to put into practice approaches for using AI and large language models to generate exercises and for feedback processes. These projects will start at CATALPA on April 1st. Regina Kasakowskij has also submitted her dissertation and plans to continue her research on feedback in teaching as a postdoc in the two new projects.