Distance learning is getting more traction with the rise of online 
learning platforms such as Coursera or DataCamp, and adopted 
rapidly due to the COVID19 crisis. The most popular approaches, 
such as exercises in a fixed learning path, are directed towards the 
entire student population and fail to take into account individual 
strengths and weaknesses. Despite its clear benefits for learners, 
one-to-one tutoring is not scalable because it is too time- and cost-
intensive. To overcome this, we develop a semantic graph-based 
approach to data-driven personalisation in education, following the 
Flemish government’s emphasis on adaptive educational systems 
towards Society 2025. Semantic graphs are a novelty in this context 
and have ideal properties to represent student behaviour and 
learning content. First, we estimate the student knowledge level by 
applying Bayesian Deep Learning on a semantic graph. Second, we 
use semantic graphs to represent educational content, applied to 
language learning and mathematics learning. The graph preserves 
important language properties and estimates the text difficulty level. 
Finally, we combine both user and content representation models 
into a single recommender system that builds the optimal learning 
path that is both challenging and engaging, concept- and context-
aware. This intelligent tutoring system will lower course development 
costs and will be suitable for a wide range of domains.