Prof. Dr. Özgür Özçep, Universität Hamburg, CHAI
Abstract:
Analogies and analogical reasoning are central topics at the intersection of various disciplines such as computer science (in particular AI), legal reasoning, cognitive science, psychology, and linguistics. This talk focuses on the model of analogical proportions understood as quaternary relations of the form a:b::c:d, read as “a is to b as c is to d.” After presenting recent general results on analogical proportions, the talk discusses their application to concrete instances in two subareas of AI: (1) mechanism design for fair resource allocation and (2) geometry-based knowledge graph embeddings.
Institution