The MOCAH team (Human Learning-Centered Computing) is specialized in Technology Enhanced Learning (TEL) with an AI background (knowledge engineering, ontologies, conversational agents, data mining). The research themes have evolved for the last 10 years, with a reuse and development of our expertise into new themes. We have been working on authoring tools and cognitive diagnosis, using mainly symbolic approach. The core of our research is proposing models that embed human knowledge in TEL systems to improve learner models, diagnosis, feedback and the systems themselves. Knowledge is gathered (1) directly from humans (experts, learners, teachers...) using knowledge engineering techniques and/or (2) from data using educational data mining techniques.

Models are associated to open computer environments such as simulations, serious games and virtual reality systems, i.e. environment with higher possibilities of interaction. Other less interactive environments like MOOCs interest us to address the issues they raise about the learner assessment, feedback and adaptation scenarios from large amount of data.

Currently MOCAH is striving to experiment in real situations with large number of students and various domains and levels. The team has obtained a chair (V. Luengo) from Sorbonne Universités in 2015 to apply their research to our university context. Additionally, MOCAH participates to 3 ANR projects, and has a contract with the DNE (Direction du Numérique Educatif). All these projects are funded until 2018.

In years to come, MOCAH is planning to reinforce its research in the following themes:

• Tracking, diagnosis and feedback taking into account data heterogeneity coming from various sources (activity from simulator, serious game, LMS, MCQ, etc.) and not only actions but also perceptions (eyetracking, haptic...). We are interested in exploring the potential of new paradigms such as connected objects (IoT), mobile learning, virtual and augmented reality, and open collaborative platforms.
• Working in merging knowledge extracted from data with the ones built from humans.
• Learning analytics methods and tools to help stakeholders, in particular teachers, to adapt and personalize learning.

This involves in particular the use of techniques from Natural Language Processing to analyze students’ interactions with teachers and among themselves to build better knowledge models and/or propose adapted activities like students’ groups.