Searcharxiv⌕ Search

arXiv subjects

Sebastian Möller

Publications and source records attributed to Sebastian Möller.

At least 19 recordsLinked to original sources
arXiv2026

Assessing Quality of Experience in Natural Language Generation of German Text

The rapid advancement of Natural Language Generation (NLG) has made the reliable evaluation of generated text increasingly critical, as these systems, such as large language models (LLMs), are now widely deployed in real-world applications. However, traditional automatic metrics fail to capture the multifaceted nature of perceived quality. In this paper, we introduce TextQ-German, a novel dataset suite for human-centered evaluation of German NLG from a Quality of Experience (QoE) perspective, covering automatic text summarization and machine translation. Through crowdsourcing studies with German speakers, we collect human quality ratings and identify relevant perceptual quality dimensions for each task. We develop automatic QoE prediction models, including transformer-based, linguistic feature-based, and hybrid approaches. Hybrid models outperform pure transformer baselines in almost all experimental settings, while linguistic features alone can approach the performance of fine-tuned language models. The dataset is extended with LLM-generated outputs annotated with overall QoE scores. Final validation on held-out sets indicates generalization to unseen data. Our work contributes a publicly accessible resource for NLG evaluation and baselines for automatic QoE prediction, providing a foundation for developing NLG systems that better align with human quality perception.

cs.CL↗
arXiv2026

3D Gaussian Splatting and Mesh-Based Digital Twins: An Exploratory Study for Virtual Reality Tourism

Digital Twins (DTs) are increasingly used for immersive experiences in virtual tourism. Virtual Reality (VR) enables remote visits to replicated locations for promotional purposes or access to fragile and rural cultural heritage sites. However, developing high-fidelity DTs of tourist destinations is costly, due to the manual creation of 3D environments. Novel 3D rendering techniques, such as 3D Gaussian splatting (3DGS), pose a promising approach to creating immersive experiences. This study investigates the user experience (UX) of a 3D-mesh-based scene and a 3DGS-based scene within a VR tourism application. In a laboratory study, 20 participants engaged with both versions and rated UX, cybersickness, presence and affect through standardized questionnaires. A custom questionnaire was created to measure the perception of the DTs. The collected data suggests that both versions were enjoyed and induced positive affect, with the Mesh version receiving good UX ratings. While the 3DGS version scored higher in terms of experienced realism, it showed clear weaknesses in pragmatic quality. Further, the results suggest that the feeling of presence could be enhanced and cybersickness reduced in both versions. Overall, the study contributes to the understanding of UX in VR tourism applications by implementing mesh-based and 3DGS-based DTs and raising important questions about the perception of realism.

cs.HC↗
arXiv2026

Large Language Models as Unified Multimodal Learners for Clinical Prediction

Electronic health records combine free-text clinical narratives with structured measurements such as vital signs, laboratory values, and comorbidities. Yet most clinical prediction systems still rely on task-specific fusion architectures, pairing dedicated encoders for each modality with learned combination mechanisms that must be re-engineered for every new task and clinical setting. We propose a simpler alternative: convert all patient data, regardless of modality, into a single natural language sequence and fine-tune a pretrained language model end-to-end, with no architectural modification for fusion. We evaluate this approach across three clinically distinct prediction tasks: in-hospital mortality on MIMIC-III, graft failure prediction using longitudinal data from a German transplant center, and emergency triage classification from ambulance records - comparing encoder-based (ModernBERT) and decoder-based (Llama 3.1, Gemma, DeepSeek-R1-Qwen, Qwen3) fine-tuning against established multimodal baselines and, for graft failure, a gradient boosting model currently used in clinical practice for post-transplant patient management. Across all three tasks, unified textual serialization matches or exceeds task-specific multimodal baselines, and outperforms the clinically deployed gradient boosting system on graft failure prediction. These results indicate that a single serialization-based paradigm, without bespoke fusion architectures, is sufficient for multimodal clinical prediction - substantially reducing system complexity while matching or exceeding specialized designs.

cs.CL↗
arXiv2026

Evidence Subspace Projection: Measuring How Much Evidence Explains Deepfake Detection in Self-Supervised Speech Models

Self-supervised learning (SSL) models are widely used as feature extractors for state-of-the-art audio deepfake detection, but it remains unclear how to directly and quantitatively connect what SSL models capture to detection decisions. To address this gap, we propose Evidence Subspace Projection, a method that represents both evidence factors (e.g., attack category, codec, gender, transmission) and authenticity labels in a shared space constructed from SSL models' neuron activation patterns. By projecting the decision vector onto each evidence subspace, we obtain a scalar ratio that quantifies the explanatory power of each evidence type. We evaluate SSL models in raw, fine-tuned, and post-trained settings on multiple datasets. The results confirm findings from established studies, validating the proposed method, and reveal new insights into model behavior.

cs.SD↗
arXiv2026

DialogPII: A multilingual dataset of synthetic dialog transcripts to detect personal information

Conversational data collected in domains such as healthcare or social sciences is a valuable resource for research and automated analysis. However, responsible data sharing requires the detection and removal of personally identifiable and sensitive information to protect individual privacy. To support the development and evaluation of automatic de-identification systems, we present DialogPII, a multilingual dataset of synthetic dialogs and speech-derived transcripts for personal information detection. DialogPII covers eight interaction scenarios (emergency calls, medical anamnesis interviews, therapy sessions, insurance communication, customer support, clinical interviews regarding an AI-supported dashboard, police reports, and group therapy discussions), 19 entity types, and 11 languages (English, Arabic, Finnish, French, German, Hindi, Italian, Polish, Portuguese, Spanish, and Turkish). Dialogs were generated semi-automatically using large language models, manually curated for plausibility and diversity, and localized to country- and city-specific contexts. All dialogs were additionally converted to speech via text-to-speech synthesis, transcribed with Whisper, and annotated through automatic projection and manual correction, yielding aligned written and speech-derived resources across all languages. We further release baseline multilingual named entity recognition models and provide technical validation through inter-annotator agreement analysis, translation quality evaluation, annotation projection assessment, and benchmark experiments with transformer-based sequence labeling models.

cs.CL↗
arXiv2026

Does Finetuning with Scientific Data Increase Hallucinations? A Multi-domain Factuality Evaluation of LLMs

Large language models (LLMs) are increasingly used to communicate and explain scientific concepts, yet their tendency to hallucinate poses significant risks in this high stakes use-case. Prior scientific hallucination evaluation work remains largely restricted to the biomedical domain, treats hallucination as a binary task, and has not examined the growing family of scientifically fine-tuned LLMs. We address these gaps with SciFactCheck, a benchmark of 2,500 prompts across five scientific domains, paired with a modular evaluation framework targeting three factuality hallucination types: unverifiability, overclaim, and attribution. Using a controlled minimal-pairing design, we evaluate 18 LLMs by comparing each scientifically fine-tuned model against its general-purpose base. Our results indicate that 1. Scientifically fine-tuned models exhibit degraded factual reliability across all hallucination types and scientific domains, and 2. Fine-tuned models are internally less confident yet linguistically more assertive. A human pilot study further reveals that current fact-checking tools show only modest agreement with expert judgments on scientific content, and that defining scientifically check-worthy claims remains contested even among human annotators. Our findings fundamentally challenge current methods of domain-specific fine-tuning for factuality and call for developing improved verification infrastructure for scientific content.

cs.CL↗
arXiv2026

Judge Circuits Explain Format-Induced Inconsistency in LLM-as-a-Judge

LLM-as-a-judge has become the dominant paradigm for grading model outputs at scale, yet the same model assigns systematically different scores when its output format changes (e.g., a 1-5 rating vs. a True/False label). Existing diagnoses of these format-induced inconsistencies stop at the input-output level. Using Position-aware Edge Attribution Patching (PEAP), we causally investigate the internal mechanism in five open-weight instruction-tuned models (Gemma-3, Qwen2.5, Llama-3.1) across five judgment tasks. We find that judgments across structured understanding and open-ended preference tasks share a sparse Latent Evaluator sub-graph in the mid-to-late multi-layer perceptrons (MLPs); zero-ablating it collapses judgment while preserving performance on our knowledge probes in architecturally modular models. By structurally decoupling abstract judging from output formatting, we provide a mechanistic account of format-induced inconsistency on the open-weight models we study: a continuous judgment signal computed in the shared trunk is mapped through fragile, format-specific terminal branches. The judgment itself can therefore be read out independently of the requested output format. Our findings imply that benchmark comparisons of judge reliability across formats partly measure the fragile formatting stage, and can understate the quality of the underlying evaluation.

cs.CL↗
arXiv2026

Fine-tuning with Hierarchical Prompting for Robust Propaganda Classification Across Annotation Schemas

Propaganda detection in social media is challenging due to noisy, short texts and low annotation agreements. We introduce a new intent-focused taxonomy of propaganda techniques and compare it against an established, higher-agreement schema. Along three dimensions (model portfolio, schema effects, and prompting strategy) we evaluate the taxonomies as a classification task with the help of four language models (GPT-4.1-nano, Phi-4 14B, Qwen2.5-14B, Qwen3-14B). Our results show that fine-tuning is essential, since it transforms weak zero-shot baselines into competitive systems and reveals methodological differences that are hidden using base models. Across schemas, the Qwen models achieve the strongest overall performance, and Phi-4 14B consistently outperforms GPT-4.1-nano. Our hierarchical prompting method (HiPP), which predicts fine-grained techniques before aggregating them, is especially beneficial after fine-tuning and on the more ambiguous, low-agreement taxonomy, while remaining competitive on the simpler schema. The HQP dataset, annotated with the new intent-based labels, provides a richer lens on propaganda's strategic goals and a challenging benchmark for future work on robust, real-world detection.

cs.CL↗
arXiv2026

From Articles to Premises: Building PrimeFacts, an Extraction Methodology and Resource for Fact-Checking Evidence

Fact-checking articles encode rich supporting evidence and reasoning, yet this evidence remains largely inaccessible to automated verification systems due to unstructured presentation. We introduce PrimeFacts, a methodology and resource for extracting fine-grained evidence from full fact-checking articles. We compile 13,106 PolitiFact articles with claims, verdicts, and all referenced sources, and we identify 49,718 in-article hyperlinks as natural anchors to pinpoint key evidence. Our framework leverages large language models (LLMs) to rewrite these anchor sentences into stand-alone, context-independent premises and investigates the extraction of additional implicit evidence. In evaluations on cross-article evidence retrieval and claim verification, the extracted premises substantially improve performance. Decontextualized evidence yields higher retrievability, achieving up to a 30 percent relative gain in Mean Reciprocal Rank over verbatim sentences, and using the evidence for verdict prediction raises Macro-F1 by 10-20 points over the baseline. These gains are consistent across different verdict granularities (2-class vs. 5-class) and model architectures. A qualitative analysis indicates that the decontextualized premises remain faithful to the original sources. Our work highlights the promise of reusing fact-checkers' evidence for automation and provides a large-scale resource of structured evidence from real-world fact-checks.

cs.CL↗
arXiv2026

Gendered Prompting and LLM Code Review: How Gender Cues in the Prompt Shape Code Quality and Evaluation

LLMs are increasingly embedded in programming workflows, from code generation to automated code review. Yet, how gendered communication styles interact with LLM-assisted programming and code review remains underexplored. We present a mixed-methods pilot study examining whether gender-related linguistic differences in prompts influence code generation outcomes and code review decisions. Across three complementary studies, we analyze (i) collected real-world coding prompts, (ii) a controlled user study, in which developers solve identical programming tasks with LLM assistance, and (iii) an LLM-based simulated evaluation framework that systematically varies gender-coded prompt styles and reviewer personas. We find that gender-related differences in prompting style are subtle but measurable, with female-authored prompts exhibiting more indirect and involved language, which does not translate into consistent gaps in functional correctness or static code quality. For LLM code review, in contrast, we observe systematic biases: on average, models approve female-authored code more, despite comparable quality. Controlled experiments show that gender-coded prompt style affect code length and maintainability, while reviewer behavior varies across models. Our findings suggest that fairness risks in LLM-assisted programming arise less from generation accuracy than from LLM evaluation, as LLMs are increasingly deployed as automated code reviewers.

cs.SE↗
arXiv2026

Retrieving Climate Change Disinformation by Narrative

Detecting climate disinformation narratives typically relies on fixed taxonomies, which do not accommodate emerging narratives. Thus, we re-frame narrative detection as a retrieval task: given a narrative's core message as a query, rank texts from a corpus by alignment with that narrative. This formulation requires no predefined label set and can accommodate emerging narratives. We repurpose three climate disinformation datasets (CARDS, Climate Obstruction, climate change subset of PolyNarrative) for retrieval evaluation and propose SpecFi, a framework that generates hypothetical documents to bridge the gap between abstract narrative descriptions and their concrete textual instantiations. SpecFi uses community summaries from graph-based community detection as few-shot examples for generation, achieving a MAP of 0.505 on CARDS without access to narrative labels. We further introduce narrative variance, an embedding-based difficulty metric, and show via partial correlation analysis that standard retrieval degrades on high-variance narratives (BM25 loses 63.4% of MAP), while SpecFi-CS remains robust (32.7% loss). Our analysis also reveals that unsupervised community summaries converge on descriptions close to expert-crafted taxonomies, suggesting that graph-based methods can surface narrative structure from unlabeled text.

cs.CL↗
arXiv2026

MultiGraSCCo: A Multilingual Anonymization Benchmark with Annotations of Personal Identifiers

Accessing sensitive patient data for machine learning is challenging due to privacy concerns. Datasets with annotations of personally identifiable information are crucial for developing and testing anonymization systems to enable safe data sharing that complies with privacy regulations. Since accessing real patient data is a bottleneck, synthetic data offers an efficient solution for data scarcity, bypassing privacy regulations that apply to real data. Moreover, neural machine translation can help to create high-quality data for low-resource languages by translating validated real or synthetic data from a high-resource language. In this work, we create a multilingual anonymization benchmark in ten languages, using a machine translation methodology that preserves the original annotations and renders names of cities and people in a culturally and contextually appropriate form in each target language. Our evaluation study with medical professionals confirms the quality of the translations, both in general and with respect to the translation and adaptation of personal information. Our benchmark with over 2,500 annotations of personal information can be used in many applications, including training annotators, validating annotations across institutions without legal complications, and helping improve the performance of automatic personal information detection. We make our benchmark and annotation guidelines available for further research.

cs.CL↗
arXiv2026

Dishonesty Tendencies in Testing Scenarios Among Students with Virtual Reality and Computer-Mediated Technology

Virtual reality (VR) systems have the potential to be an innovation in the field of e-learning. Starting with fully functional e-classes, VR technologies can be used to build entire e-campuses. The power of VR is that it allows for stronger contact with students than computer-mediated technology. Deceptive behaviour, both verbal and nonverbal, refers to intentional activities designed to deceive others. Students often engage in dishonest practices to make progress. Whether it is cheating on an exam, copying another student's essay, or inflating their GPA, the motivation for cheating is rarely simply a lack of preparation. Even though some may see academic dishonesty as an asset, the reality is that it can have major consequences. This poster demonstrates the findings from a study of students' deceitful behaviour during a test in VR and in real-life situations. For this user study, 22 volunteers were invited to participate, with each experiment involving exactly two participants and the examiner present in the room. Students were invited to take two tests: one in VR and one on a laptop. Their goal was to score as many points as possible by simulating a real-world online exam. Participants were requested to complete questionnaires during and after each experiment, which assisted in collecting additional data for this study. The results indicate that the amount of cheating that happened in VR and on a laptop was exactly the same.

cs.HC↗
arXiv2026

Integrating Virtual and Augmented Reality into Public Education: Opportunities and Challenges in Language Learning

Virtual Reality (VR) and Augmented Reality (AR) are emerging as transformative tools in education, offering new possibilities for engagement and immersion. This paper explores their potential in language learning within public education, focusing on their ability to enhance traditional schooling methods and address existing educational gaps. The integration of VR and AR in schools, however, is not without challenges, including usability, technical barriers, and the alignment of these technologies with existing curricula. Drawing on two empirical studies, this work investigates the opportunities and challenges of VR- and AR-assisted language learning and proposes strategies for their effective implementation in the public sector. The findings show that VR increases motivation and immersion but has an unclear impact on vocabulary retention, with technical limitations and cognitive overload identified as key challenges. AR enhances contextual learning and accessibility but faces usability constraints and limited personalization. To facilitate effective adoption, this paper recommends improving interface design, reducing cognitive load, increasing adaptability, and ensuring adequate infrastructure and teacher training. Overcoming these barriers will enable a more effective integration of immersive technologies in language education.

cs.HC↗
arXiv2026

Influence of Interactivity in Shaping User Experience and Social Acceptance of Mobile XR

This study investigates the impact of the Degree of Interactivity on User Experience (UX) and social acceptability (SA) in Mobile Augmented Reality (MAR) applications. As AR technologies become more prevalent, understanding how varying levels of interactivity influence both user perception and social dynamics is crucial for their design and adoption. Two commercially available MAR applications, IKEA and Virtlo, which differ significantly in their interactivity levels, were used to conduct a user study. The study examines how body movements required for interaction with AR content affect both UX and SA, shedding light on users' comfort levels and potential social barriers in public settings. The findings suggest a complex relationship between interactivity, perceived usability, and social considerations, emphasizing the need for a balanced design approach. This research provides valuable insights into the development of future AR applications by addressing not only usability but also the broader social implications of AR interactions. By integrating social acceptability into traditional UX evaluations, this study highlights its significance in ensuring the seamless integration of AR technologies into everyday environments.

cs.HC↗
arXiv2026

Evaluating Parkinson's Disease Detection in Anonymized Speech: A Performance and Acoustic Analysis

Automatic detection of Parkinson's disease (PD) from speech is a promising non-invasive diagnostic tool, but it raises significant privacy concerns. Speaker anonymization mitigates these risks, but it may suppress the pathological information necessary for PD detection. We assess the trade-off between privacy and PD detection for two anonymizers (STT-TTS and kNN-VC) using two Spanish datasets. STT-TTS provides better privacy but severely degrades PD detection by eradicating prosodic information. kNN-VC preserves macro-prosodic features such as duration and F0 contours, achieving F1 scores only 3-7\% lower than original baselines, demonstrating that privacy-preserving PD detection is viable when using appropriate anonymization. Finally, an acoustic distortion analysis characterizes specific weaknesses in kNN-VC, offering insights for designing anonymizers that better preserve PD information.

cs.SD↗
arXiv2026

Exploring the Effect of Heights and User Stance on User Experience in Extended Reality Climbing

Virtual environments (VEs) are increasingly used for immersive experiences, training simulations, and entertainment, yet factors such as height perception and user stance can significantly influence user experience (UX). Height perception in VEs plays a crucial role in shaping UX, particularly in immersive applications such as climbing simulations. This study investigates the effects of height in various VEs and examines how user stance, sitting or standing, impacts immersion, perceived height, and motion sickness. A user study was conducted with 25 participants who played through five randomized climbing scenarios, ranging from indoor climbing gyms to outdoor cityscapes and mountainous terrains. Participants' UX was assessed using standardized questionnaires, including the IPQ for general presence, spatial presence, involvement, and experienced realism, as well as the SSQ to evaluate motion sickness symptoms such as nausea, oculomotor strain, and disorientation. Results indicate that seated participants experienced slightly higher immersion but were also more susceptible to motion sickness compared to those standing. While standing participants maintained consistent scores across different environments, seated participants reported increased immersion and discomfort as the VEs became larger, more physically demanding, and visually complex.

cs.HC↗
arXiv2026

Too Immersive for the Field? Addressing Safety Risks in Extended Reality User Studies

Extended Reality (XR) technologies are increasingly tested outside the lab, in homes, schools, and public spaces. While this shift enables more realistic user insights, it also introduces safety challenges that are often overlooked. Physical risks, psychological distress, and accessibility issues can be increased in field studies and unsupervised testing, such as at home or crowdsourced trials. Without clear instructions, safety decisions are left to individual researchers, raising questions of responsibility and consistency. This position paper outlines key safety risks in XR user testing beyond the lab and calls for practical strategies that are needed to help researchers run XR studies in a safe and inclusive way across different environments.

cs.HC↗