Papers by Christin Seifert
Give Me the Facts! A Survey on Factual Knowledge Probing in Pre-trained Language Models (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Pre-trained language models are trained on vast unlabeled data, rich in world knowledge. |
| Approach: | They propose a categorization scheme for factual probing methods based on how inputs, outputs and probed PLMs are adapted . they synthesize insights about knowledge retention and prompt optimization in PLM models and analyze obstacles to adopting them as knowledge bases . |
| Outcome: | The proposed method synthesizes insights about knowledge retention and prompt optimization in PLMs, analyzes obstacles to adopting them as knowledge bases and outline directions for future work. |
Persuasion Tokens for Editing Factual Knowledge in LLMs (2026.eacl-short)
Copied to clipboard
| Challenge: | In-context knowledge editing (IKE) relies on fact-specific demonstrations which consume significant context window space. |
| Approach: | They introduce persuasion tokens (P-Tokens) which replicate the effect of IKE demonstrations and allow efficient knowledge editing without requiring fact-specific demonstrations. |
| Outcome: | The proposed tokens perform comparable to and often exceed IKE on two editing datasets and three LLMs and increase the number of tokens increases performance. |
InfoLossQA: Characterizing and Recovering Information Loss in Text Simplification (2024.acl-long)
Copied to clipboard
Jan Trienes, Sebastian Joseph, Jörg Schlötterer, Christin Seifert, Kyle Lo, Wei Xu, Byron Wallace, Junyi Jessy Li
| Challenge: | Text simplification aims to make technical texts more accessible to laypeople but often results in deletion of information and vagueness. |
| Approach: | They propose a framework to characterize and recover simplification-induced information loss in form of question-and-answer (QA) pairs. |
| Outcome: | The proposed framework characterizes and recovers simplification-induced information loss in form of question-and-answer (QA) pairs. |
How to Make LLMs Forget: On Reversing In-Context Knowledge Edits (2025.naacl-long)
Copied to clipboard
| Challenge: | In-context knowledge editing (IKE) is an efficient and efficient knowledge editing method (Zheng et al., 2022b; Gangadhar and Stratos, 2024) it can be misused to manipulate responses opaquely, e.g., insert misinformation or offensive content. |
| Approach: | They propose to detect and reverse IKE-edits using only the top-10 output probabilities of the next token, even in a black-box setting. |
| Outcome: | The proposed method can be detected with high accuracy even in a black-box setting, achieving over 80% accuracy in recovering original, unedited outputs across multiple LLMs. |
Parallel Universes, Parallel Languages: A Comprehensive Study on LLM-based Multilingual Counterfactual Example Generation (2026.acl-long)
Copied to clipboard
Qianli Wang, Van Bach Nguyen, Yihong Liu, Fedor Splitt, Nils Feldhus, Christin Seifert, Hinrich Schuetze, Sebastian Möller, Vera Schmitt
| Challenge: | Large language models excel at generating English counterfactuals but their effectiveness in generating multilingual counterfacts remains unclear. |
| Approach: | They conduct automatic evaluations on both directly generated and derived counterfactuals in six languages and find that cross-lingual perturbations follow common strategic principles. |
| Outcome: | The proposed models show that translation-based counterfactuals offer higher validity than their directly generated counterparts, but still fall short of matching the quality of the original English counterf actuals. |
One Mask to Rule Them All: On Hidden Facts after Editing and How to Find Them (2026.findings-acl)
Copied to clipboard
| Challenge: | Knowledge editing methods such as ROME and MEMIT update factual associations by modifying MLP weights. |
| Approach: | They propose to use a mask to reverse edits by eliminating overattention in later layers . they also show that injecting the mask during editing drops editing success from 98% to 38% . |
| Outcome: | The proposed method reverses edits by eliminating overattention in later layers and drops editing success from 98% to 38%. |
How Do LLMs Generate Contrastive Sentiments? A Mechanistic Perspective (2026.eacl-long)
Copied to clipboard
| Challenge: | Despite extensive research, the mechanisms underlying LLMs' abilities remain poorly understood. |
| Approach: | They propose and validate a mechanistic intervention that transforms the sentiment of a text from positive to negative while making minimal edits. |
| Outcome: | The proposed intervention increases sentiment flip rate without sacrificing minimal changes to text content. |
Behavioral Analysis of Information Salience in Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models excel at text summarization, but the exact notion of salience remains unclear. |
| Approach: | They propose a framework to derive and investigate information salience in Large Language Models (LLMs) using length-controlled summarization as a behavioral probe into the content selection process. |
| Outcome: | The proposed framework derives a proxy for how models prioritize information in large language models. |
Has this Fact been Edited? Detecting Knowledge Edits in Language Models (2025.naacl-long)
Copied to clipboard
| Challenge: | Knowledge editing methods (KEs) can update language models’ obsolete or inaccurate knowledge learned from pre-training. |
| Approach: | They propose to detect knowledge edits in language models by using four KEs, two large language models and two datasets to classify the knowledge as unedited (based on pre-training) and edited (based upon subsequent editing). |
| Outcome: | The proposed method detects whether an output is based on edited knowledge or first-hand knowledge from pre-training. |
Marcel: A Lightweight and Open-Source Conversational Agent for University Student Support (2025.emnlp-demos)
Copied to clipboard
Jan Trienes, Anastasiia Derzhanskaia, Roland Schwarzkopf, Markus Mühling, Jörg Schlötterer, Christin Seifert
| Challenge: | Existing systems that provide contextually relevant information are difficult to deploy in a university setting . a number of universities are developing or using chatbots to support prospective students . |
| Approach: | They propose a conversational agent called Marcel that uses retrieval-augmented generation to provide contextually relevant information. |
| Outcome: | The proposed system is designed to provide fast and personalized responses while reducing workload. |
The Queen of England is not England’s Queen: On the Lack of Factual Coherency in PLMs (2024.findings-eacl)
Copied to clipboard
| Challenge: | Existing work evaluated how often PLMs can correctly predict a subject and a relation . previous work focused on evaluating how much PLM know, but this study focused on the internal state of knowledge inside them. |
| Approach: | They examine how often PLMs can correctly predict a subject and a relation . they also examine how knowledge inside PLM is embodied in the internal state . |
| Outcome: | The proposed model improves on the accuracy of the evidence paragraphs and manually written prompts. |
Comprehensive Study on German Language Models for Clinical and Biomedical Text Understanding (2024.lrec-main)
Copied to clipboard
Ahmad Idrissi-Yaghir, Amin Dada, Henning Schäfer, Kamyar Arzideh, Giulia Baldini, Jan Trienes, Max Hasin, Jeanette Bewersdorff, Cynthia S. Schmidt, Marie Bauer, Kaleb E. Smith, Jiang Bian, Yonghui Wu, Jörg Schlötterer, Torsten Zesch, Peter A. Horn, Christin Seifert, Felix Nensa, Jens Kleesiek, Christoph M. Friedrich
| Challenge: | Pre-trained language models can struggle in specialized domains such as medicine . existing generalpurpose pre-tried models can be used and refined through further pre-training on domainspecific unlabeled data. |
| Approach: | They pre-trained German medical language models on 2.4B tokens from translated public data and 3B token of German clinical data. |
| Outcome: | The proposed models outperform clinical models on various downstream tasks in germany . the authors show that continuous pre-training can match or exceed clinical models trained from scratch . |
LLMs for Generating and Evaluating Counterfactuals: A Comprehensive Study (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown remarkable performance in NLP tasks, but their efficacy in generating high-quality CFs remains uncertain. |
| Approach: | They compare LLMs' ability to generate CFs that flip the original label and human CF's. |
| Outcome: | The proposed models generate fluent CFs, but struggle to keep the induced changes minimal. |