Computational analysis of group trajectories in collaborative learning

September 26, 2018 – ,


Research in the field of adaptive and intelligent systems for collaborative learning support aims to capture and model information of group activity and use this information to monitor and support students and teachers. Using computational methods, the complexity of collaborative processes shall be transformed in understandable and useful representations. These representations can be used (1) to get insights about the requirements for productive collaborative learning scenarios and (2) to design interventions which support learners and/or teachers in the learning process. A possible intervention is the reformation of a group of learners.
In this theses two methods are to be developed and evaluated which automatically segment group discussions in a forum/discussion board in micro-activities and classifies the micro activities. The first method uses known approaches from Natural Language Processing and supervised machine learning. As second method, deep learning approaches and word resp. document vectors should be used.


  1. Familiarization with the related work
  2. Familiarization with the coding scheme and the existing corpus of already labeled group discussions
  3. Design and implementation of a method for segmentation of group discussions (NLP and ML)
  4. Design and implementation of a method for classification of these segments (NLP and ML)
  5. Design and implementation of an alternative method for segmentation and segment classification (Deep Learning / word resp. document vectors)
  6. Comparative evaluation of the different methods based on an existing corpus


  • Interest in Educational Technologies
  • Experience in Natural Language Processing
  • Experience in Machine Learning / Deep Learning



download corresponding tendering


Research Area(s): Knowledge & Educational Technologies

Tutor: Christoph Rensing,

Open Theses