Sprachtechnologie und Multilingualität
Der Forschungsbereich „Sprachtechnologie und Multilingualität“ (MLT) in Saarbrücken wird von Prof. Dr. Josef van Genabith geleitet.
MLT befasst sich mit folgenden Schwerpunktthemen: Maschinelle Übersetzung (MT), Frage-Antwort- und Informationsextraktions-Systeme (QAIE), „Talking Robots“ (TR) sowie Erklärbare Sprachverarbeitung (E&E).
- MT-Gruppe: maschinelle Übersetzung zwischen Texten und von Text in Gebärdensprache;
- QAIE-Gruppe: Hochskalierbare Text-Analyse- und darauf aufbauend Frage-Antwort- und Informationsextraktions-Systeme;
- TR-Gruppe: Dialogverarbeitung für Mensch-Roboter Kollaboration und für die Unterstützung von Teamarbeit. Grundlegende Dialogtechnologien sowie Anwendungen in z.B. Katastrophenschutz und Gesundheit;
- E&E-Gruppe: Effiziente und erklärbare NLP-Modelle.
Wir arbeiten mit maschinellem Lernen und wissensbasierten Ansätzen in Forschung und Entwicklung. Neben der Grundlagenforschung ist MLT stark in Auftragsforschung und Industrieprojekten engagiert.
An der Universität des Saarlandes sind Mitglieder der MLT-Gruppe in leitender Funktion im SFB 1102 (Information Density and Linguistic Encoding, Teilprojekt B6), einem DFG-Projekt zu multi-modalen Interfaces für das Post-Editing von MÜ-Ergebnissen (MMPE), sowie dem Europäischen „Erasmus Mundus“-Master-Programm in „Language and Communication Technologies“ (LCT) engagiert.
Vorträge bei MLT-Meetings
2024/25
| Referent | Zugehörigkeit | Titel | Datum |
|---|---|---|---|
| Daniil Gurgurov | DFKI | External Knowledge-Driven Adaptation of Multilingual Language Models for Low-Resource Language Enhancement (MSc Thesis defense) | May 27, 2025 |
| Tatiana Anikina | DFKI | Reverse Probing: Evaluating Knowledge Transfer via Task Embeddings for Coreference Resolution | May 20, 2025 |
| Steffen Castle | DFKI | Entity Linking using LLMs for Automated Product Carbon Footprint Estimation | May 13, 2025 |
| Alex Duchnowski | Saarland University | EHOP: A Dataset of Everyday NP-Hard Optimization Problems | May 6, 2025 |
| Qianli Wang | DFKI | FitCF: A Framework for Automatic Feature Importance-guided Counterfactual Example Generation | April 29, 2025 |
| Gabriele Sarti | University of Groningen | QE4PE: Word-level Quality Estimation for Human Post-Editing | April 15, 2025 |
| Sayed Muddashir Hossain | DFKI | Automatic Recognition of Psychodynamic Conflicts from Semi- structured Interviews with Large Language Models (MSc Thesis defense) | April 8, 2025 |
| Ruben Härle | TU Darmstadt | SCAR: Sparse Conditioned Autoencoders for Concept Detection and Steering in LLMs | April 1, 2025 |
| Christian Schuler | MTACR: Multilingual Text As Corpus Repository - Improving Machine Translation of Low-Resource Languages | March 18, 2025 | |
| Christopher Schröder | ScaDS.AI and Leipzig University | Self-Training for Sample-Efficient Active Learning for Text Classification with Pre-Trained Language Models | March 3, 2025 |
| Günter Neumann | DFKI | AtLas: AI based Natural Language Processing of Low Quality and Multilingual Data in Defence Applications with User Adaptation | February 18, 2025 |
| Daniil Gurgurov | DFKI | GrEmLIn: A Repository of Green Baseline Embeddings for 87 Low-Resource Languages Injected with Multilingual Graph Knowledge | February 11, 2025 |
| Cennet Oguz | DFKI | MuJo: Multimodal Joint Feature Space Learning for Human Activity Recognition | February 4, 2025 |
| Vitalii Hirak | Universität des Saarlandes | Explaining Machine Translation Difficulty in the Age of Massively Multilingual Models: A Study of the Impact of Fine-Grained Typological Features and Beam Size on State-of-the-Art NMT (MSc Thesis Defense) | January 28, 2025 |
| Prajvi Saxena | DFKI | Streamlining LLMs: Adaptive Knowledge Distillation for Tailored Language Models. | January 21, 2025 |
| Julian Schlenker | DFKI | On the Efficacy of Language Adapters for Cross-lingual Transfer in English-centric LLMs (MSc Thesis Defense) | January 14, 2025 |
| Bangyao Tang | DFKI | Analysis of Calibrated Confident Text Classification (MSc Thesis Defense) | December 10, 2024 |
| Gabriele Merlin | Max Planck Institute for Software Systems | Language models and brains align due to more than next-word prediction and word-level information | November 19, 2024 |
| Xinting Huang | Universität des Saarlandes | InversionView: A General-Purpose Method for Reading Information from Neural Activations | November 12, 2024 |
| Vagrant Gautam | Universität des Saarlandes | Understanding "Democratization" in NLP and ML Research | November 5, 2024 |
| Ivan Vykopal | Kempelen Institute of Intelligent Technologies (KInIT) | Soft Prompts for Language Transfer | October 22, 2024 |
| Johannes Ruf, David Riebschläger | DFKI | Workshop: Efficient Hosting of Large Language Models with GH200 and vLLM | October 15, 2024 |
| Robert Belanec | Kempelen Institute of Intelligent Technologies (KInIT) | Multi-Task Soft-Prompt Transfer for LLMs via Task Prompt Vectors | October 8, 2024 |
| Tatiana Anikina, Arne Binder | DFKI | DFKI-MLST at DialAM-2024 Shared Task: System Description (+ more experimental results) | October 1, 2024 |
| Shakib Yazdani | DFKI | Continual Learning in Multilingual Sign Language Translation | September 24, 2024 |
| Yasser Hamidullah | DFKI | Sign Language Translation with Sentence Embedding Supervision | September 17, 2024 |
| Miaoran Zhang | Universität des Saarlandes | The Impact of Demonstrations on Multilingual In-Context Learning: A Multidimensional Analysis | September 10, 2024 |
| Vagrant Gautam, Marius Mosbach | Universität des Saarlandes | From Insights to Actions: The Impact of Interpretability and Analysis Research on NLP | September 3, 2024 |
| Jenny Kunz | Linköpings Universiteit | Properties and Challenges of LLM-Generated Explanations | August 27, 2024 |
| Daniil Gurgurov | DFKI | Adapting Multilingual LLMs to Low-Resource Languages with Knowledge Graphs via Adapters | August 20, 2024 |
| Noon Pokaratsiri | DFKI | Towards Understanding Attention-based Reasoning through Graph Structures in Medical Codes Classification | August 6, 2024 |
| Yusser Al Ghussin | DFKI | Semantic Search with DAM | July 30, 2024 |
| Martin Courtois | DFKI | Symmetric Dot-Product Attention for Efficient Training of BERT Language Models | July 23, 2024 |
| Koel Dutta Chowdhury, Maria Kunilovskaya | Universität des Saarlandes | Symmetric Dot-Product Attention for Efficient Training of BERT Language ModelsMitigating Translationese with GPT-4: Strategies and Performance | July 16, 2024 |
| Johannes Leveling | Fraunhofer IAIS | Training Large Language Models in OpenGPT-X | July 9, 2024 |
| Ivana Beňová | Kempelen Institute of Intelligent Technologies (KInIT) | Beyond Image-Text Matching: Probing Understanding in Multimodal Transformers | July 2, 2024 |
| Michal Gregor, Ivan Vykopal | Kempelen Institute of Intelligent Technologies (KInIT) | Introducing KInIT and Our Work in NLP | June 18, 2024 |
| Rafael Tolosana Calasanz | Universidad Zaragoza | The AI4Europe Reproducibility Initiative & The AI on Demand Platform | June 11, 2024 |
| Kurt Micallef, Kurt Abela | University of Malta | NLG & MT Developments for Maltese | June 4, 2024 |
| Philipp Müller | DFKI | Multi-modal Conversation Analysis | May 28, 2024 |
| Tatiana Anikina, Nils Feldhus | DFKI | LLMCheckup: Conversational Examination of Large Language Models via Interpretability Tools and Self-Explanations | May 21, 2024 |
| Koel Dutta Chowdhury | Universität des Saarlandes | A Representation Learning Based Approach To The Study of Translationese | May 14, 2024 |
| Yuekun Yao | Universität des Saarlandes | Simple and effective data augmentation for compositional generalization | May 7, 2024 |
| Yasser Hamidullah | DFKI | Gloss-free Spatio-Temporal Sign Language representation | April 30, 2024 |
| Jan Niehues | Karlsruhe Institute of Technology (KIT) | Multilingual Support in the Era of Foundation Models | April 23, 2024 |
| Fabrizio Nunnari | DFKI | Representation and Synthesis of Sign Language: the MultiModal Signstream concept | April 16, 2024 |
| Stalin Varanasi, Umer Butt | DFKI | AIden: A Student Assistant for navigating course material | April 9, 2024 |
| Doreen Osmelak | Universität des Saarlandes | Code-Switching in German-English: The Role of Shared Lexical Items (MSc Thesis Defense) | March 12, 2024 |
Leitung
Leitung:
Prof. Dr. Josef van Genabith
josef.van_genabith@dfki.de
Tel.: +49 681 85775 5287
Stellvertretung:
Dr. Simon Ostermann
simon.ostermann@dfki.de
Tel.: +49 681 85775 5298
Kontakt
Sekretariat:
Tel.: +49 681 85775 5292
mlt-sek@dfki.de
Deutsches Forschungszentrum für
Künstliche Intelligenz GmbH (DFKI)
Campus D3 2
Stuhlsatzenhausweg 3
66123 Saarbrücken
Deutschland
News aus dem Forschungsbereich
Alle MLT News
