Seminar by Prof. Ana Ozaki (Online)

Description Logics (DLs) are formalisms designed for representing knowledge in the format of an ontology and performing automated reasoning. The most prominent applications of DLs are for data integration and query answering. There are several repositories with DL ontologies expressing knowledge in various domains (medicine, biology, music, etc). In a DL ontology, knowledge is expressed in a declarative manner (e.g., "Cats are Mammals"). This explicit representation together with automated reasoning tools allow for precise control of which information is considered to be true in a knowledge-based system, what is the source of that information, which parts of the ontology are relevant for retrieving that information, and so on. Knowledge can also be uncertain. For instance, consider an image recognition system that associates a probability value p1 for an image to contain a cat, and another probability p2 value for it to contain a dog). If the system contains the information that "Cats are Mammals" and "Dogs are Mammals" then it can infer that the probability that the image contains a mammal is at least max(p1,p2). The synergy between knowledge and uncertainty can result in applications that are both reliable and practical. In this talk, I will give a gentle introduction to DLs and ways of combining it with uncertainty reasoning.

Bio: Prof. Ana Ozaki completed her BSc and MSc in Computer Science at the University of Brasilia and she completed her PhD degree in Computer Science at the University of Liverpool in 2016. After positions in TU Dresden, the Free University of Bozen-Bolzano and at the University of Bergen, she is currently an associate professor at the University of Oslo. She works on knowledge representation, reasoning, learning theory and their intersections with machine learning.