Agenda

Speaker
Rocco Tripodi, University of Bologna

Join Zoom Meeting
https://unive.zoom.us/s/81788857035

Abstract:
Abstract Meaning Representation (AMR) is a popular formalism of natural language that represents the meaning of a sentence as a rooted, directed, and usually acyclic graph where the nodes represent concepts and the edges represent semantic relations among concepts. It is agnostic about how to derive meanings from strings and for this reason, it was used successfully on different tasks, including human-robot interaction, visual question answering, information extraction, and machine translation. This feature makes AMR particularly suited also to encode the meaning of sentences in different languages with the same graph structure. In recent years different models have been proposed for AMR parsing, including transition-based approaches that transform syntactic structures into semantic structures and sequence-to-graph transduction approaches that use the attention mechanism to identify potential nodes and relations in text. However, the results on AMR parsing were boosted recently using generative language models such as BART and T5. This trend gave rise to the development of new parsers such as SPRING, SGL, and StructBart trained end-to-end to decode linearized graph structures starting from plain text. This lecture will give an overview of the AMR task, its evaluation, and its models.

Bio Sketch:
Rocco Tripodi is assistant professor at the University of Bologna. He completed his Ph.D. in Computer Science at Ca' Foscari University of Venice in 2016 with a thesis titled "Evolutionary Game Theoretic Models for Natural Language Processing,” He has worked as researcher in various laboratories including Sapienza NLP at Sapienza University of Rome and European Centre for Living Technology (ECLT) in Venice and under different European and industrial projects including Hume-Nash Machines, ODYCCEUS, MOUSSE and Polifonia. Rocco Tripodi has a multidisciplinary attitude. He has worked in both Computer Science and Linguistics departments and groups. His research interests are in the areas of machine learning and NLP with a focus on lexical semantics and natural language understanding. He is particularly interested in learning models based on transfer learning principles and in the design, learning and evolution of linguistic communication systems.