CDD Researcher (M/F) in Large Language Models (LLMs) for natural language argumentation

CNRS; France; Sur place; EUR 40 000 - 60 000; Plein temps

Résumé du poste

A major research institution in France is seeking a post-doctoral researcher to advance studies in artificial intelligence focused on natural language argumentation. The role involves developing models using large language models and collaborating with interdisciplinary teams. Candidates must hold a Ph.D. in relevant fields and have strong skills in AI frameworks and natural language processing. This position offers a unique opportunity to contribute to cutting-edge research.

Qualifications

• Demonstrated experience in NLP, AI, or related areas through published research.
• Specific experience in computational argumentation or argument mining will be a plus.

Responsabilités

• Design new models for argumentation using large language models.
• Validate proposed models on standard datasets.
• Disseminate research results in conferences and journals.

Connaissances

Familiarity with frameworks like TensorFlow

Experience with tools for AI and NLP

Skills in data collection and analysis

Ability to evaluate algorithms for argumentation

Formation

Completed Ph.D. in Computer Science or related field

Outils

TensorFlow

PyTorch

Hugging Face Transformers

Description du poste

Organisation/Company CNRS Department Institut de Recherche en Informatique de Toulouse Research Field Computer science Mathematics » Algorithms Researcher Profile First Stage Researcher (R1) Country France Application Deadline 16 Dec 2025 - 23:59 (UTC) Type of Contract Temporary Job Status Full-time Hours Per Week 35 Offer Starting Date 1 Feb 2026 Is the job funded through the EU Research Framework Programme? Not funded by a EU programme Is the Job related to staff position within a Research Infrastructure? No

Offer Description

The researcher position is envisioned within the ANITI HUCAD project which aims to develop Artificial Intelligence (AI) systems fostering human-machine collaboration for effective deliberations. It aims to enhance the argumentation capabilities of stakeholders by suggesting relevant arguments, retrieved from the web, and facilitating a good grasp of a debate thanks to an automatically generated and structured synthesis of the most salient arguments. Computational argumentation (CA) is a sub-field of AI that studies argumentation theory, namely the theoretical foundations and applications leading to developments including logics for representing and reasoning about arguments, and axiomatically-grounded methods of argument evaluation. Natural language argumentation is a young research area at the crossroads of natural language processing (NLP) and information retrieval (IR) that focuses on identifying, retrieving, and generating arguments (and counter-arguments) present in texts and dialogues, as well as classifying relations between arguments (support, attack), and building argumentative essays and summaries. The most salient applications are intelligent personal assistants and argument search engines. However, despite the salient developments in natural language argumentation —including argument mining (AM) and argument retrieval (AR)— there still critical limitations to current natural language argumentation methods and models among which —but not limited to— the lack of reasoning capabilities to output consistent arguments that convey formal logical relations beyond the surface level of arguments, their limited generalizability regarding argument structures (e.g., nested arguments) and types (e.g., blocking arguments), the lack of objective evaluation of argument quality using normative methods.

Through this researcher position, we specifically aim to explore the synergistic benefits of CA and natural language argumentation to design theoretically grounded models of natural language argumentation by leveraging the capabilities of Large Language Models (LLMs) as compelling means for natural language understanding and reasoning.

Language Models (LMs) have emerged as the backbone of many tasks involving reasoning and language understanding in AI, IR and NLP. In argumentation, the literature reveals that they are effective for many argument retrieval and mining tasks (e.g., argument relation classification). More recently, LLMs that are inherently known as “soft “reasoners under zero-shot and few-shot settings, have allowed impressive improvements over a large set of language and decision-making tasks. However, they still suffer from critical drawbacks, including the lack of truthfulness and logical reasoning abilities in complex reasoning tasks. Leveraging LLM for argumentation is still at an infancy stage (e.g., preliminary works on argument relation classification and argumentation-based explanation ), leaving a number of the above-cited challenges unsolved such as the logical inconsistency of a set of output arguments and their low-quality given the lack of objective evaluation.

The main objective of this post-doctoral position is to participate in the design of dedicated models for human-like argumentation that fully leverage the theoretical developments in computation argumentation (CA), information retrieval (IR), and natural language processing (NLP) using recent advances in large language models (LLMs). The candidate is expected to:

• Design new axiomatically grounded end-to-end LLM-based models for natural language argumentation through tailored fine-tuning and prompting strategies;
• Probe and validate the proposed models on standard datasets using both objective evaluation and data-driven empirical evaluation;
• Disseminate the scientific results of the project in high-quality conference venues and journals in AI, IR and NLP.

Toulouse, often referred to as "La Ville Rose" due to its distinctive terracotta brick architecture, stands out as a vibrant international city that blends rich cultural heritage with cutting-edge innovation. Nestled in the heart of the Occitanie region in southwestern France, Toulouse boasts proximity to the picturesque Pyrenees mountains, offering residents and visitors alike easy access to stunning natural landscapes and outdoor activities. The city is a global aerospace hub, home to the headquarters of Airbus, which attracts a diverse, skilled workforce from around the world. Toulouse is also a prominent center for research and education, with renowned institutions like the Toulouse Capitole University, University of Toulouse-Jean Jaurès, Tolouse University, National Polytechnic Institute of Toulouse, Institut National des Sciences Appliquées de Toulouse, Higher Institute of Aeronautics and Space, to mention a few, fostering a dynamic environment for academic and scientific advancement. Its lively cultural scene, coupled with a high quality of life, makes Toulouse an attractive destination for international professionals, students, and tourists.

The Artificial and Natural Intelligence Toulouse Institute (ANITI) is a major AI scientific hub established in Toulouse. The challenge is to make Toulouse one of the world leaders in artificial intelligence in research, education, innovation and economic development, attracting top-tier researchers and professionals from around the world. ANITI's presence bolsters the city's reputation as a hub for technological advancement, complementing its status as the headquarters of Airbus and its rich aerospace heritage. ANITI has three main missions: scientific research, training and economic development related to AI. ANITI aims to develop a new generation of artificial intelligence called hybrid AI, integrating data driven learning techniques and symbolic or mathematical models that permit us to express constraints and to carry out logical reasoning. ANITI also has ambitious aims with regard to education and industrial development.

Technical Skills

• Familiarity with frameworks like TensorFlow, PyTorch, or similar is essential.
• Experience with tools and libraries for AI and NLP, such as Hugging Face Transformers or similar.
• Skills in data collection, cleaning, preprocessing, and analysis, including working with large datasets.
• Ability to design, implement, and evaluate algorithms for computational argumentation and related tasks will be a plus.

Soft Skills

• Excellent written and verbal communication skills, with the ability to present research findings clearly and effectively.
• Ability to work collaboratively in a research team, including interdisciplinary collaborations.

Research Experience

• Demonstrated experience in NLP, AI, or related areas through published research papers in top-tier journals or conferences.
• Specific experience or research focus in computational argumentation, argument mining, argument retrieval, or discourse analysis using large language models will be more.

Educational Qualifications

• A completed completed Ph.D. in a relevant field such as Computer Science, Computational Linguistics, Artificial Intelligence, or a closely related discipline.