Description logic ontologies have been used to represent the relevant knowledge of a domain of interest in a formal and machine-processable format. Such knowledge can be applied to constrain the space of hypotheses in learning tasks, to integrate data coming from multiple sources, to support query answering, among others. Understanding how to describe domain knowledge in a concise and interpretable way is a fundamental challenge in artificial intelligence. In this presentation, we will look into some examples of expressions and see which ones we can and cannot represent within classical description logic. We will also highlight some approaches in machine learning and data mining that have been applied to (semi)-automate the process of building description logic ontologies.

Ana Ozaki recently joined the machine learning group at the Universityof Bergen as an  associate professor. Ozaki is Brazilian. She obtained her bachelor's and master's degrees at the University of Brasilia and her Ph.D. degree at the University of Liverpool, United Kingdom.

She has been working on knowledge representation and computational learning theory. Ozaki is interested in the formalisation of the learning phenomenon so that questions involving learnability, complexity, and reducibility can be systematically investigated and understood in the field of machine learning. Her research focuses on learning logical theories formulated in description logic and related formalisms for knowledge representation. She has also worked on query answering and other reasoning tasks for description logic ontologies.