| Date: | 2023, July 11 |
| Time: | 11:30 a. m. |
| Place: | G29-018 |
| Author: | Glauer, Martin |
| Title: | Knowledge and Learning: Synergies between Ontologies and Machine Learning |
Ontologies are highly valuable for representing knowledge due to their rich semantics, enabling a thorough description of a particular domain that is understandable by humans. They can also be utilized to ensure consistency in existing knowledge and infer missing data. However, the creation of ontologies is typically a laborious and costly, manual process, requiring expertise from multiple domain experts. On the other hand, machine learning exhibits impressive predictive performance, often achieving remarkable results with minimal human involvement. However, modern machine learning approaches rely heavily on vast amounts of data that are not always available or easily collectible.
In my thesis, I aim to bridge these two domains. By harnessing the power of machine learning, we can expedite the development of ontologies, resulting in more efficient creation of a useful representation of knowledge. Likewise, by leveraging the domain-specific knowledge and robust semantics provided by existing ontologies, we can guide machine-learning approaches. To illustrate these concepts, I will present my work centred around the ontology for Chemicals of Biological Interest (ChEBI).