One paper by Qianli Wang and Nils Feldhus has been accepted to the Findings Track at the 63rd Annual Meeting of the Association for Computational Linguistics 2025 (ACL 2025). In the paper, they introduce ZeroCF, a faithful approach for leveraging important words derived from feature attribution methods to generate counterfactual examples in a zero-shot setting.

One paper originating from research in the TRAILS project has been accepted to the 10th Workshop on Representation Learning for NLP (RepL4NLP 2025), co-located with NAACL 2025. In the paper, we reimagine classical probing to evaluate knowledge transfer from simple source to more complex target tasks.

One paper originating from research in the TRAILS project has been accepted to the Main Track at the 31st International Conference on Computational Linguistics 2025 (COLING 2025). In the paper, we present CROSS-REFINE, a generator-critic framework that enhances natural language explanations by refining initial outputs using feedback from a second LLM, outperforming SELF-REFINE and working effectively even with less powerful models.

One paper from researchers in the DFKI-NLP group has been accepted as a Findings paper at the 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP 2024). EMNLP will take place November 12-16 in Miami, Florida.

Two papers from DFKI NLP researchers have been accepted at the 17th International Natural Language Generation Conference (INLG 2024) that will take place September 23-27 in Tokyo, Japan. One paper presents a case study on using large language models to produce customer-friendly help page contents from more technical text, and includes a text quality evaluation by experienced editors.

Counterfactual examples are widely used in natural language processing (NLP) as valuable data to improve models, and in explainable artificial intelligence (XAI) to understand model behavior. The automated generation of counterfactual examples remains a challenging task even for large language models (LLMs), despite their impressive performance on many tasks. In this paper, we first introduce ZeroCF, a faithful approach for leveraging important words derived from feature attribution methods to generate counterfactual examples in a zero-shot setting. Second, we present a new framework, FitCF, which further verifies aforementioned counterfactuals by label flip verification and then inserts them as demonstrations for few-shot prompting, outperforming two state-of-the-art baselines. Through ablation studies, we identify the importance of each of FitCF’s core components in improving the quality of counterfactuals, as assessed through flip rate, perplexity, and similarity measures. Furthermore, we show the effectiveness of LIME and Integrated Gradients as backbone attribution methods for FitCF and find that the number of demonstrations has the largest effect on performance. Finally, we reveal a strong correlation between the faithfulness of feature attribution scores and the quality of generated counterfactuals.

In this work, we reimagine classical probing to evaluate knowledge transfer from simple source to more complex target tasks. Instead of probing frozen representations from a complex source task on diverse simple target probing tasks (as usually done in probing), we explore the effectiveness of embeddings from multiple simple source tasks on a single target task. We select coreference resolution, a linguistically complex problem requiring contextual understanding, as focus target task, and test the usefulness of embeddings from comparably simpler tasks tasks such as paraphrase detection, named entity recognition, and relation extraction. Through systematic experiments, we evaluate the impact of individual and combined task embeddings. Our findings reveal that task embeddings vary significantly in utility for coreference resolution, with semantic similarity tasks (e.g., paraphrase detection) proving most beneficial. Additionally, representations from intermediate layers of fine-tuned models often outperform those from final layers. Combining embeddings from multiple tasks consistently improves performance, with attention-based aggregation yielding substantial gains. These insights shed light on relationships between task-specific representations and their adaptability to complex downstream tasks, encouraging further exploration of embedding-level task transfer.

Growing concerns about climate change and sustainability are driving manufacturers to take significant steps toward reducing their carbon footprints. For these manufacturers, a first step towards this goal is to identify the environmental impact of the individual components of their products. We propose a system leveraging large language models (LLMs) to automatically map components from manufacturer Bills of Materials (BOMs) to Life Cycle Assessment (LCA) database entries by using LLMs to expand on available component information. Our approach reduces the need for manual data processing, paving the way for more accessible sustainability practices.

Natural language explanations (NLEs) are vital for elucidating the reasoning behind large language model (LLM) decisions. Many techniques have been developed to generate NLEs using LLMs. However, like humans, LLMs might not always produce optimal NLEs on first attempt. Inspired by human learning processes, we introduce Cross-Refine, which employs role modeling by deploying two LLMs as generator and critic, respectively. The generator outputs a first NLE and then refines this initial explanation using feedback and suggestions provided by the critic. Cross-Refine does not require any supervised training data or additional training. We validate Cross-Refine across three NLP tasks using three state-of-the-art open-source LLMs through automatic and human evaluation. We select Self-Refine (Madaan et al., 2023) as the baseline, which only utilizes self-feedback to refine the explanations. Our findings from automatic evaluation and a user study indicate that Cross-Refine outperforms Self-Refine. Meanwhile, Cross-Refine can perform effectively with less powerful LLMs, whereas Self-Refine only yields strong results with ChatGPT. Additionally, we conduct an ablation study to assess the importance of feedback and suggestions. Both of them play an important role in refining explanations. We further evaluate Cross-Refine on a bilingual dataset in English and German.

Clinical narratives provide comprehensive patient information. Achieving interoperability involves mapping relevant details to standardized medical vocabularies. Typically, natural language processing divides this task into named entity recognition (NER) and medical concept normalization (MCN). State-of-the-art results require supervised setups with abundant training data. However, the limited availability of annotated data due to sensitivity and time constraints poses challenges. This study addressed the need for unsupervised medical concept annotation (MCA) to overcome these limitations and support the creation of annotated datasets.