***********************************************************************************************************************************
Workshop 4 AS
Workshop on Advances in AMR and Semantic Analysis
4AS@TALN2025 - Marseille - June, 30th 2025

****************************
https://team.inria.fr/semagramme/fr/atelier-4-as/
****************************

Many researchers have dreamed of creating general artificial intelligence, and to do so have imagined exploiting the semantics of language (parsing and generation). Today, we’re getting closer to this possibility when we see the ability of models to grasp complex problems. But none of these systems really exploits semantic analysis, which remains a fertile field of research in terms of formalisms, architectures/models and publications. Although imperfect, a formalism that attempts to bridge these different issues, AMR, has been the focus of much effort.
This workshop aims to bring together the various teams working on language semantics, with three objectives.

• Take stock of the different approaches to exploiting language semantics, its various formalisms (DRS, AMR, Yarn, …) and their uses (more frugal AI models, answering problems where semantics has clear advantages over LLMs, …). The focus will be on AMR, its strengths, weaknesses, potential successors and future developments.

• Identify and position the various resources useful for semantic analysis. While the development of Propbank in French does not appear to be a winning strategy, the development of a word alignment from French to Propbank, along the lines of VerbNet and Semlink, would be interesting. The question of annotated corpora is also wide open.

• Exchange different points of view on the interest of pursuing this research in the face of the capabilities of very large language models (LLMs), the best way of structuring this research (collaborative projects, for example) and communicating to a wide audience.

--------------------
Themes
--------------------
The workshop invites papers that address one or more of the following themes:
• Syntax-Semantics Interface;
• Resources for semantics;
• Expanding or coupling semantic resources with LLMs ;
• Designing and annotating semantic representations;
• Comparison of semantic representation frameworks;
• Automatic text generation from meaning representations;
• Strengths and weaknesses of existing semantic representations;
• Use of semantic representations in real-life applications;
• Application of semantic representations and multilingualism ;
• Multimodality in meaning representations;
• The relationship between symbolic representations of meaning and distributed semantic representations;
• Formal properties of meaning representations;

--------------------
Submission
--------------------
The expected length of submissions is 4 pages, plus one page for camera ready versions.
Submissions must follow the style sheet below and be submitted as PDF files via the EasyChair system.

Translations of previously accepted submissions to larger conferences are accepted.

The submission system is a 4AS track on the easychair site of the main conference: https://easychair.org/my/conference?conf=coriataln2025
Style sheets are common to TALN, CORIA, RECITAL and RJCRI.
An Overleaf model is available here: Feuilles de style CORIA-TALN 2025, see on the website.

--------------------
Date
--------------------
• Submission dead-line: April 22th, 2025
• Notification: May 7th, 2025
• Camera Ready version: May 14th, 2025
Given the time constraints, no extension will be possible.

--------------------
Organizers
--------------------
• Maxime Amblard (Loria, Université de Lorraine)
• Maria Boritchev (Telecom Paris Tech)
• Bruno Guillaume (Inria)
• Johannes Heinecke (Orange)
• Frédéric Herledan (Orange)