UCCA: A Computational Approach to Cross-Linguistic Semantic Representation

Explore a comprehensive lecture on UCCA (Universal Conceptual Cognitive Annotation), a groundbreaking approach to structural semantic representation in natural language processing. Delve into the UCCA scheme and its typologically motivated characterization of abstract semantic structure. Learn about transition-based UCCA parsing and experiments leveraging multi-task learning with other formalisms like AMR and SDP. Discover UCCA's applications in text-to-text generation tasks such as machine translation and text simplification, as well as their evaluation methods. Gain insights from Omri Abend, a faculty member at the Hebrew University, as he presents this innovative framework for cross-linguistic semantic analysis and representation.