Semantic Analysis and Procedural Language Understanding in the Era of Large Language Models

Abstract: 

In this talk, I will first introduce the field of semantics and the task of semantic analysis, a.k.a, semantic parsing from a multilingual perspective. In particular, we will first discuss the layers of meaning, from morphology to pragmatics, and then define the scope of semantics as a field. I will then discuss the current benchmarks and datasets spanning different meaning representations, such as sentence-level frame representations (e.g., PropBank, VerbNet, FrameNet), semantic trees (e.g., AMRs), first-order logic (FOL) and Discourse Representation Theory (DRS) which allows for document-level representation. I will continue with other popular benchmarks such as ATIS, where the task is to parse text into SQL queries; and code generation, which aims to convert text into code pieces in various programming languages (most popularly Python and shell scripts). Finally, I will define the broader related problem, named procedural language understanding, and cover the most recent studies that employ large language models to understand text, such as ProPara, Alchemy, and Recipes. I will conclude with where we stand with the latest LLM-based techniques and the challenges that still need to be addressed.

Bio: 

Dr. Gözde Gül Şahin is an Assistant Prof. at Koç University and a KUIS AI Fellow since February 2022. Previously, she was a postdoctoral researcher in the Ubiquitous Knowledge Processing (UKP) Lab at the Technical University of Darmstadt, Germany. Her research spans the fields of linguistics and machine learning, in particular semantics, multilingual representations and large language models. She completed her PhD studies in Istanbul Technical University (İTÜ) Computer Engineering department in 2018. She was a visiting researcher at the Institute for Language, Cognition and Computation (ILCC) of the University of Edinburgh in 2017. Before her Ph.D., she received her Masters and Bachelor degrees from Sabancı University in 2011 and İTÜ in 2009, respectively. She regularly serves as a PC member for *ACL conferences and is a co-organizer for the Workshop on Multilingual Representation Learning (MRL). Her research on NLP has been funded by Tübitak 2232, and 2236 grant programs that are granted to outstanding young principal investigators.

Open Data Science

 

 

 

Open Data Science
One Broadway
Cambridge, MA 02142
info@odsc.com

Privacy Settings
We use cookies to enhance your experience while using our website. If you are using our Services via a browser you can restrict, block or remove cookies through your web browser settings. We also use content and scripts from third parties that may use tracking technologies. You can selectively provide your consent below to allow such third party embeds. For complete information about the cookies we use, data we collect and how we process them, please check our Privacy Policy
Youtube
Consent to display content from Youtube
Vimeo
Consent to display content from Vimeo
Google Maps
Consent to display content from Google