Thesis title: Learning Concept Descriptions from Examples

Supervisor: Patrick Koopmann (

The aim of this project is develop learning methods that can be used to aid the development of ontologies that are based on OWL or description logics. In particular, the problem to be solved can be stated as follows: given a dataset with some objects marked as positive and negative examples, find a logical description (using an ontology language) that describes all of the positive and none of the negative examples. While this sounds like a classical machine learning problem, the aim is to use logic-based techniques to solve this learning problem. For this, you will implement an further investigate a new idea for learning concept descriptions, which will explore recent tools for non-classical reasoning.

The supervisor will give an introduction to the topic and the proposed idea at the beginning of the project. You can contact the supervisor if you would like to have more information on this project or would like to discuss it in more detail in person.