Bahar İlGen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Bahar İlGen has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Georges Hattab
Center for Artificial Intelligence in Public Health Research (ZKI-PH) Robert Koch Institute Berlin Germany
7 shared publications - Akshat Dubey
PH) Center for Artificial Intelligence in Public Health Research (ZKI Robert Koch Institute Nordufer 20 13353 Berlin Germany
3 shared publications - Aleksandar Anžel
PH) Center for Artificial Intelligence in Public Health Research (ZKI Robert Koch Institute Nordufer 20 13353 Berlin Germany
3 shared publications - Antonia Pilic
Robert Koch Institute, Immunization Unit, Berlin, Germany
1 shared publication - Chisom Ezekannagha
Center for Artificial Intelligence in Public Health Research, Robert Koch Institute, Berlin, Germany
1 shared publication - Thomas Harder
Immunization Unit, Robert Koch Institute, Berlin, Germany
1 shared publication
Latest Publications
- Open Access
- 9 September 2026
- Published by Elsevier in Machine Learning with Applications
- Vol. 25
UbiQVision: Spatial Dempster–Shafer fusion of XAI attributions for reliable deep vision ensembles
- 9 September 2026
- Published by Elsevier in Machine Learning with Applications
- Vol. 25
Open AccessUbiQVision: Spatial Dempster–Shafer fusion of XAI attributions for reliable deep vision ensembles
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Computer Graphics and Applications
- Vol. 46 (1) , 89-98
Visualizing the Chain of Thought in Large Language Models
This Visualization Viewpoints article explores how visualization helps uncover and communicate the internal chain-of-thought trajectories and generative pathways of large language models (LLMs) in reasoning tasks. As LLMs become increasingly powerful and widespread, a key challenge is understanding how their reasoning dynamics unfold, particularly in natural language processing (NLP) applications. Their outputs may appear coherent, yet the multistep inference pathways behind them remain largely hidden. We argue that visualization offers an effective avenue to illuminate these internal mechanisms. Moving beyond attention weights or token saliency, we advocate for richer visual tools that expose model uncertainty, highlight alternative reasoning paths, and reveal what the model omits or overlooks. We discuss examples, such as prompt trajectory visualizations, counterfactual response maps, and semantic drift flows, to illustrate how these techniques foster trust, identify failure modes, and support deeper human interaction with these systems. In doing so, visualizing the chain of thought in LLMs lays critical groundwork for transparent, interpretable, and truly collaborative human-AI reasoning.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Computer Graphics and Applications
- Vol. 46 (1) , 89-98
Open AccessVisualizing the Chain of Thought in Large Language Models
This Visualization Viewpoints article explores how visualization helps uncover and communicate the internal chain-of-thought trajectories and generative pathways of large language models (LLMs) in reasoning tasks. As LLMs become increasingly powerful and widespread, a key challenge is understanding how their reasoning dynamics unfold, particularly in natural language processing (NLP) applications. Their outputs may appear coherent, yet the multistep inference pathways behind them remain largely hidden. We argue that visualization offers an effective avenue to illuminate these internal mechanisms. Moving beyond attention weights or token saliency, we advocate for richer visual tools that expose model uncertainty, highlight alternative reasoning paths, and reveal what the model omits or overlooks. We discuss examples, such as prompt trajectory visualizations, counterfactual response maps, and semantic drift flows, to illustrate how these techniques foster trust, identify failure modes, and support deeper human interaction with these systems. In doing so, visualizing the chain of thought in LLMs lays critical groundwork for transparent, interpretable, and truly collaborative human-AI reasoning.
- Open Access
- 13 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 218-224
A Lexical Simplification Framework for Turkish
- 13 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 218-224
Open AccessA Lexical Simplification Framework for Turkish
- Open Access
- 13 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 24-31
Persona Adaptable Strategies Make Large Language Models Tractable
- 13 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 24-31
Open AccessPersona Adaptable Strategies Make Large Language Models Tractable
- Open Access
- 13 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 431-438
Visualizing Evidence through Natural Language Interaction for Public Health Interventions
- 13 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 431-438
Open AccessVisualizing Evidence through Natural Language Interaction for Public Health Interventions
- Open Access
- 15 December 2023
- Published by Association for Computing Machinery (ACM)
- p. 234-239
Pre-Training to Identify Immunization-Related Entities from Systematic Reviews
Entity recognition from semi or unstructured systematic reviews is one of the most essential processes for evidence-based decision-making systems. The task involves collecting information from diverse studies concerning PICO (Population, Intervention, Comparison, and Outcomes) elements with additional domain-related information using named entity recognition (NER) as it is the fundamental task for extracting the structured data. In this study, we create an adapted immunization-related dataset and evaluate its performance in the extraction of relevant entities from systematic reviews. We conducted experiments to investigate several models for entity recognition performance using language models pre-trained in the biomedical domain. Our results suggest that PubMedBERT and BertNER results are superior to the other models, and the immunization-related entities can be successfully recognized with a 76% F1 score and 92% accuracy.
- 15 December 2023
- Published by Association for Computing Machinery (ACM)
- p. 234-239
Open AccessPre-Training to Identify Immunization-Related Entities from Systematic Reviews
Entity recognition from semi or unstructured systematic reviews is one of the most essential processes for evidence-based decision-making systems. The task involves collecting information from diverse studies concerning PICO (Population, Intervention, Comparison, and Outcomes) elements with additional domain-related information using named entity recognition (NER) as it is the fundamental task for extracting the structured data. In this study, we create an adapted immunization-related dataset and evaluate its performance in the extraction of relevant entities from systematic reviews. We conducted experiments to investigate several models for entity recognition performance using language models pre-trained in the biomedical domain. Our results suggest that PubMedBERT and BertNER results are superior to the other models, and the immunization-related entities can be successfully recognized with a 76% F1 score and 92% accuracy.