Curriculum Vitae

Download PDF version

Current Position

Postdoctoral Researcher
IDSIA (Dalle Molle Institute for Artificial Intelligence) - SUPSI Lugano, Switzerland
February 2025 - Present
Supervisor: Fabio Rinaldi

Education

Ph.D. in Language Sciences
Université Sorbonne Nouvelle, Paris, France
February 2025
Thesis: Négation et Modèles de Langue Pré-entraînés (Negation and Pretrained Language Models)
Supervisors: Marie Candito (MCF-HDR, Université Paris Cité) and Pascal Amsili (Professor, Université Sorbonne Nouvelle)
Research Labs: Lattice (UMR8094) and LLF (UMR 7110)

Master’s Degree (Maîtrise) in Computer Science
Specialization: Artificial Intelligence
Université de Montréal, Québec, Canada
2021
Research Lab: RALI (Recherche Appliquée en Linguistique Informatique)
Thesis: Méthodologies pour la détection de diachronies sémantiques et leurs impacts
Supervisor: Philippe Langlais

Bachelor’s Degree (Licence) in Computer Science
Specialization: DANT (Développement d’Applications Nouvelles Technologies)
Sorbonne Université, Paris, France
2019

Research Interests

My research is situated at the intersection of theoretical linguistics and natural language processing, within the field of computational linguistics. I approach this discipline as a branch of language sciences, where computational methods serve to interrogate and illuminate linguistic theories.

Primary Research Line: Understanding how linguistic theory can inform and improve neural language model architectures and training procedures.

Current Research Topics:

• Negation processing in pretrained language models
• Hallucination detection and mitigation in RAG systems
• Multilingual capabilities and language family influences in LLMs
• Model interpretability and compositional semantics
• Distributional semantics and contextual embeddings

Methodological Approaches:

• Formal semantics and compositionality
• Distributional semantics and vector representations
• Syntax-semantics interface
• Language model evaluation and probing

Teaching Experience

Total: 406 hours equivalent TD

Université Sorbonne Nouvelle, Paris, France (116 HTD)

Computational Linguistics (Master 1)
36 hours CM | 2022-2024 (3 years)

• Co-created course introducing NLP to language sciences students
• Developed pedagogical materials (slides, exercises, linguistic examples)
• Evaluation: midterm exam, final exam, group presentation of recent NLP paper

Exploration of Annotated Corpora (Licence 3)
48 hours TD | 2021-2023 (2 years)

• Theoretical and practical training in corpus linguistics
• Formal foundations (automata, regular expressions)
• Hands-on project using Unitex/GramLab for linguistic phenomenon identification
• Coordinated multiple TD groups

Introduction to NLP (Master 2 ENEAD - Distance Learning)
14 hours TD | 2022-2023

• Adapted in-person course for distance learning
• Created comprehensive course manual with theory, exercises, and solutions
• Produced video capsules for key concepts
• Managed Q&A forum and produced targeted supplementary videos

Université de Rouen Normandie, Rouen, France (10 HTD)

Natural Language Processing (Master 2)
10 hours TD | 2023-2024

• Advanced NLP course for students with strong CS background
• Focus on recent neural approaches and practical Python implementation
• Hands-on: parsing, pretrained models, NLP API integration

Université de Montréal, Montréal, Canada (290 HTD)

Introduction to Artificial Intelligence (Licence 3)
2021 (January-June)

Object-Oriented Programming (Licence-Master)
2020-2021 (September-January)

Supervision

Master’s Research Internships

Viola Gullace (Scuola Normale Superiore, Pisa & ENS-PSL, Paris)
November 2023 - June 2024
Topic: Impact of context polarity on encoding in contextual embeddings

Viola Gullace (Università di Pisa & ENS-PSL, Paris)
February - May 2023
Topic: How Do Pretrained Language Models Deal with Negation? The case of Italian
Outcome: paper at CLiC-it 2024

Milena Nedeljkovic (École Polytechnique)
January - April 2023
Topic: Impact of negation on the distribution of lexical items

Thesis Committee Member

Master 2 Committees:

• April 2024: Viola Gullace, “How Do Pretrained Language Models Deal with Negation? The case of Italian”
• June 2023: Delphine Nguyen-Durandet, “Détection automatique et études statistiques de la polarité en anglais”
• September 2022: Haiyan QIU, “An analysis of negation in Natural Language Inference benchmarks”

Master 1 Committee:

• June 2024: Master Cognitive Science, ENS-PSL (Licence 3 internship)

Professional Experience

Data Scientist and Analyst
Thales LAS France, Rungis, France
September 2018 - August 2019

Selected Publications

Conference Papers (Peer-Reviewed)

1. Polarity inversion operators in PLM
   David Kletz, Pascal Amsili, and Marie Candito
   29th Conference on Computational Natural Language Learning (CoNLL), pages 312-322, Vienna, Austria
   DOI: 10.18653/v1/2025.conll-1.20

2. Swushroomsia at SemEval-2025 Task 3: Probing LLMs’ Collective Intelligence for Multilingual Hallucination Detection
   Sandra Mitrović*, Joseph Cornelius*, David Kletz*, Ljiljana Dolamic, and Fabio Rinaldi
   19th International Workshop on Semantic Evaluation (SemEval-2025)

3. Assessing RAG System Capabilities on Financial Documents
   Oscar Lithgow-Serrano, David Kletz, et al.
   10th Workshop on Financial Technology and Natural Language Processing (FinNLP)
   DOI: 10.18653/v1/2025.finnlp-2.9

4. The Self-Contained Negation Test Set
   David Kletz, Pascal Amsili, and Marie Candito
   6th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP, pages 212-221, Singapore
   DOI: 10.18653/v1/2023.blackboxnlp-1.16

5. Probing structural constraints of negation in Pretrained Language Models
   David Kletz, Marie Candito, and Pascal Amsili
   24th Nordic Conference on Computational Linguistics (NoDaLiDa), pages 541-554, Tórshavn, Faroe Islands

6. EvoSem: A database of polysemous cognate sets
   Mathieu Dehouck, Alex François, Siva Kalyan, Martial Pastor, and David Kletz
   4th Workshop on Computational Approaches to Historical Language Change (LChange), pages 66-75, Singapore
   DOI: 10.18653/v1/2023.lchange-1.7

Full publication list

Selected Invited Talks

• “Negation encoding in PLMs: Syntactic constraints and consequences” - ColingLab Seminar, Scuola Normale Superiore, Pisa, Italy (2024)
• “Probing structural constraints of negation in Pretrained Language Models” - NLP Group, Saarland University, Saarbrücken, Germany (2023)

Contact Information

Email: david.kletz@supsi.ch

Office: IDSIA, East Campus USI-SUPSI Via la Santa 1. CH-6962 Lugano-Viganello

Website: https://davkletz.github.io

GitHub: github.com/davkletz

Last updated: January 2025