arXiv subjects

Claudia Wagner

Publications and source records attributed to Claudia Wagner.

At least 19 recordsLinked to original sources

arXiv2026

When Persona Attributes Improve Population Alignment in Large Language Models

Large Language Models (LLMs) are increasingly used to predict the responses of human participants in survey panels. Towards that goal, persona prompting has recently emerged as a technique to inform and align large pretrained language models. Persona prompting refers to the practice of using short textual descriptions of 'personas' in prompts to steer the LLM's generations. Personas describe individuals through different attributes such as their socio-demographics, attitudes, or behaviors, with the aim of aligning LLMs to produce responses that correlate with the corresponding human responses. Yet, recent work has produced mixed and partly conflicting results of persona prompting without clear patterns of success and failure. Among the few consistent findings is that the selection of persona attributes matters, and that using more attributes does not necessarily lead to better performance. It remains unclear how different attribute selection methods perform and how to choose among them. In this paper, we propose that observed human response variation of a survey question is a potential explanation for the mixed performance observed so far. In addition, we compare the performance of persona prompting associated with different methods for selecting persona attributes. We evaluate these methods on four different (general) social surveys across two countries, six LLMs, and twenty prediction tasks per survey. Our work helps to identify when persona prompting can be expected to be useful in survey prediction tasks, and provides new insights on the effectiveness of different attribute selection methods for LLM-based survey prediction using persona prompting.

cs.CL↗

arXiv2026

Hybrid Panels: Toward Human-AI Collaboration in Survey Research

Large-scale population surveys are essential for generating robust social and scientific insights, yet they face significant challenges, including declining response rates, increasing data collection costs, long delays between data collection and data provision, and the risk of nonresponse bias. Advances in artificial intelligence (AI) have opened up new opportunities for AI-supported survey infrastructures where the goal is to overcome these challenges without limiting the data quality. A promising AI-enabled survey infrastructure for which we build a first pilot is a hybrid panel. A hybrid panel is a longitudinal AI-enabled survey which allows to iteratively improve the alignment between large language models (LLMs) and the population they aim to simulate and use the errors to inform the design and implementation of the next survey wave (e.g., inform the participant recruitment, assignment of questions to participants). It incorporates both human participants and LLMs as fundamental elements of its design. In this research note, we introduce the concept of a hybrid panel by providing a definition and outlining an overarching framework, spanning data collection to data validation. We detail results from a first pilot study to illustrate (open) challenges that we identify for hybrid panels.

cs.CL↗

arXiv2026

From Emotion to Expression: Theoretical Foundations and Resources for Fear Speech

Few forces rival fear in their ability to mobilize societies, distort communication, and reshape collective behavior. In computational linguistics, fear is primarily studied as an emotion, but not as a distinct form of speech. Fear speech content is widespread and growing, and often outperforms hate-speech content in reach and engagement because it appears "civiler" and evades moderation. Yet the computational study of fear speech remains fragmented and under-resourced. This can be understood by recognizing that fear speech is a phenomenon shaped by contributions from multiple disciplines. In this paper, we bridge cross-disciplinary perspectives by comparing theories of fear from Psychology, Political science, Communication science, and Linguistics. Building on this, we review existing definitions. We follow up with a survey of datasets from related research areas and propose a taxonomy that consolidates different dimensions of fear for studying fear speech. By reviewing current datasets and defining core concepts, our work offers both theoretical and practical guidance for creating datasets and advancing fear speech research.

cs.CL↗

arXiv2025

German General Social Survey Personas: A Survey-Derived Persona Prompt Collection for Population-Aligned LLM Studies

The use of Large Language Models (LLMs) for simulating human perspectives via persona prompting is gaining traction in computational social science. However, well-curated, empirically grounded persona collections remain scarce, limiting the accuracy and representativeness of such simulations. Here, we introduce the German General Social Survey Personas (GGSS Personas) collection, a comprehensive and representative persona prompt collection built from the German General Social Survey (ALLBUS). The GGSS Personas and their persona prompts are designed to be easily plugged into prompts for all types of LLMs and tasks, steering models to generate responses aligned with the underlying German population. We evaluate GGSS Personas by prompting various LLMs to simulate survey response distributions across diverse topics, demonstrating that GGSS Personas-guided LLMs outperform state-of-the-art classifiers, particularly under data scarcity. Furthermore, we analyze how the representativity and attribute selection within persona prompts affect alignment with population responses. Our findings suggest that GGSS Personas provide a potentially valuable resource for research on LLM-based social simulations that enables more systematic explorations of population-aligned persona prompting in NLP and social science research.

cs.CL↗

arXiv2025

Beyond the Explicit: A Bilingual Dataset for Dehumanization Detection in Social Media

Digital dehumanization, although a critical issue, remains largely overlooked within the field of computational linguistics and Natural Language Processing. The prevailing approach in current research concentrating primarily on a single aspect of dehumanization that identifies overtly negative statements as its core marker. This focus, while crucial for understanding harmful online communications, inadequately addresses the broader spectrum of dehumanization. Specifically, it overlooks the subtler forms of dehumanization that, despite not being overtly offensive, still perpetuate harmful biases against marginalized groups in online interactions. These subtler forms can insidiously reinforce negative stereotypes and biases without explicit offensiveness, making them harder to detect yet equally damaging. Recognizing this gap, we use different sampling methods to collect a theory-informed bilingual dataset from Twitter and Reddit. Using crowdworkers and experts to annotate 16,000 instances on a document- and span-level, we show that our dataset covers the different dimensions of dehumanization. This dataset serves as both a training resource for machine learning models and a benchmark for evaluating future dehumanization detection techniques. To demonstrate its effectiveness, we fine-tune ML models on this dataset, achieving performance that surpasses state-of-the-art models in zero and few-shot in-context settings.

cs.CL↗

arXiv2025

Learning from Convenience Samples: A Case Study on Fine-Tuning LLMs for Survey Non-response in the German Longitudinal Election Study

Survey researchers face two key challenges: the rising costs of probability samples and missing data (e.g., non-response or attrition), which can undermine inference and increase the use of convenience samples. Recent work explores using large language models (LLMs) to simulate respondents via persona-based prompts, often without labeled data. We study a more practical setting where partial survey responses exist: we fine-tune LLMs on available data to impute self-reported vote choice under both random and systematic nonresponse, using the German Longitudinal Election Study. We compare zero-shot prompting and supervised fine-tuning against tabular classifiers (e.g., CatBoost) and test how different convenience samples (e.g., students) used for fine-tuning affect generalization. Our results show that when data are missing completely at random, fine-tuned LLMs match tabular classifiers but outperform zero-shot approaches. When only biased convenience samples are available, fine-tuning small (3B to 8B) open-source LLMs can recover both individual-level predictions and population-level distributions more accurately than zero-shot and often better than tabular methods. This suggests fine-tuned LLMs offer a promising strategy for researchers working with non-probability samples or systematic missingness, and may enable new survey designs requiring only easily accessible subpopulations.

cs.CY↗

arXiv2025

Facts are Harder Than Opinions -- A Multilingual, Comparative Analysis of LLM-Based Fact-Checking Reliability

The proliferation of misinformation necessitates scalable, automated fact-checking solutions. Yet, current benchmarks often overlook multilingual and topical diversity. This paper introduces a novel, dynamically extensible data set that includes 61,514 claims in multiple languages and topics, extending existing datasets up to 2024. Through a comprehensive evaluation of five prominent Large Language Models (LLMs), including GPT-4o, GPT-3.5 Turbo, LLaMA 3.1, and Mixtral 8x7B, we identify significant performance gaps between different languages and topics. While overall GPT-4o achieves the highest accuracy, it declines to classify 43% of claims. Across all models, factual-sounding claims are misclassified more often than opinions, revealing a key vulnerability. These findings underscore the need for caution and highlight challenges in deploying LLM-based fact-checking systems at scale.

cs.CY↗

arXiv2024

Fair Pairs: Fairness-Aware Ranking Recovery from Pairwise Comparisons

Pairwise comparisons based on human judgements are an effective method for determining rankings of items or individuals. However, as human biases perpetuate from pairwise comparisons to recovered rankings, they affect algorithmic decision making. In this paper, we introduce the problem of fairness-aware ranking recovery from pairwise comparisons. We propose a group-conditioned accuracy measure which quantifies fairness of rankings recovered from pairwise comparisons. We evaluate the impact of state-of-the-art ranking recovery algorithms and sampling approaches on accuracy and fairness of the recovered rankings, using synthetic and empirical data. Our results show that Fairness-Aware PageRank and GNNRank with FA*IR post-processing effectively mitigate existing biases in pairwise comparisons and improve the overall accuracy of recovered rankings. We highlight limitations and strengths of different approaches, and provide a Python package to facilitate replication and future work on fair ranking recovery from pairwise comparisons.

cs.CY↗

arXiv2024

Sexism Detection on a Data Diet

There is an increase in the proliferation of online hate commensurate with the rise in the usage of social media. In response, there is also a significant advancement in the creation of automated tools aimed at identifying harmful text content using approaches grounded in Natural Language Processing and Deep Learning. Although it is known that training Deep Learning models require a substantial amount of annotated data, recent line of work suggests that models trained on specific subsets of the data still retain performance comparable to the model that was trained on the full dataset. In this work, we show how we can leverage influence scores to estimate the importance of a data point while training a model and designing a pruning strategy applied to the case of sexism detection. We evaluate the model performance trained on data pruned with different pruning strategies on three out-of-domain datasets and find, that in accordance with other work a large fraction of instances can be removed without significant performance drop. However, we also discover that the strategies for pruning data, previously successful in Natural Language Inference tasks, do not readily apply to the detection of harmful content and instead amplify the already prevalent class imbalance even more, leading in the worst-case to a complete absence of the hateful class.

cs.CL↗

arXiv2024

The Unseen Targets of Hate -- A Systematic Review of Hateful Communication Datasets

Machine learning (ML)-based content moderation tools are essential to keep online spaces free from hateful communication. Yet, ML tools can only be as capable as the quality of the data they are trained on allows them. While there is increasing evidence that they underperform in detecting hateful communications directed towards specific identities and may discriminate against them, we know surprisingly little about the provenance of such bias. To fill this gap, we present a systematic review of the datasets for the automated detection of hateful communication introduced over the past decade, and unpack the quality of the datasets in terms of the identities that they embody: those of the targets of hateful communication that the data curators focused on, as well as those unintentionally included in the datasets. We find, overall, a skewed representation of selected target identities and mismatches between the targets that research conceptualizes and ultimately includes in datasets. Yet, by contextualizing these findings in the language and location of origin of the datasets, we highlight a positive trend towards the broadening and diversification of this research space.

cs.CL↗

arXiv2023

Individual and gender inequality in computer science: A career study of cohorts from 1970 to 2000

Inequality prevails in science. Individual inequality means that most perish quickly and only a few are successful, while gender inequality implies that there are differences in achievements for women and men. Using large-scale bibliographic data and following a computational approach, we study the evolution of individual and gender inequality for cohorts from 1970 to 2000 in the whole field of computer science as it grows and becomes a team-based science. We find that individual inequality in productivity (publications) increases over a scholar's career but is historically invariant, while individual inequality in impact (citations), albeit larger, is stable across cohorts and careers. Gender inequality prevails regarding productivity, but there is no evidence for differences in impact. The Matthew Effect is shown to accumulate advantages to early achievements and to become stronger over the decades, indicating the rise of a "publish or perish" imperative. Only some authors manage to reap the benefits that publishing in teams promises. The Matthew Effect then amplifies initial differences and propagates the gender gap. Women continue to fall behind because they continue to be at a higher risk of dropping out for reasons that have nothing to do with early-career achievements or social support. Our findings suggest that mentoring programs for women to improve their social-networking skills can help to reduce gender inequality.

cs.CY↗

arXiv2023

People Make Better Edits: Measuring the Efficacy of LLM-Generated Counterfactually Augmented Data for Harmful Language Detection

NLP models are used in a variety of critical social computing tasks, such as detecting sexist, racist, or otherwise hateful content. Therefore, it is imperative that these models are robust to spurious features. Past work has attempted to tackle such spurious features using training data augmentation, including Counterfactually Augmented Data (CADs). CADs introduce minimal changes to existing training data points and flip their labels; training on them may reduce model dependency on spurious features. However, manually generating CADs can be time-consuming and expensive. Hence in this work, we assess if this task can be automated using generative NLP models. We automatically generate CADs using Polyjuice, ChatGPT, and Flan-T5, and evaluate their usefulness in improving model robustness compared to manually-generated CADs. By testing both model performance on multiple out-of-domain test sets and individual data point efficacy, our results show that while manual CADs are still the most effective, CADs generated by ChatGPT come a close second. One key reason for the lower performance of automated methods is that the changes they introduce are often insufficient to flip the original label.

cs.CL↗

arXiv2023

ValiText -- a unified validation framework for computational text-based measures of social constructs

Guidance on how to validate computational text-based measures of social constructs is fragmented. While researchers generally acknowledge the importance of validating text-based measures, they often lack a shared vocabulary and a unified framework to do so. This paper introduces ValiText, a new validation framework designed to assist scholars in validly measuring social constructs in textual data. The framework is built on a conceptual foundation of validity in the social sciences, strengthened by an empirical review of validation practices in the social sciences and consultations with experts. Ultimately, ValiText prescribes researchers to demonstrate three types of validation evidence: substantive evidence (outlining the theoretical underpinning of the measure), structural evidence (examining the properties of the text model and its output) and external evidence (testing for how the measure relates to independent information). The framework is further supplemented by a checklist of validation steps, offering practical guidance in the form of documentation sheets that guide researchers in the validation process.

cs.CL↗

arXiv2023

Computational Reproducibility in Computational Social Science

Replication crises have shaken the scientific landscape during the last decade. As potential solutions, open science practices were heavily discussed and have been implemented with varying success in different disciplines. We argue that computational-x disciplines such as computational social science, are also susceptible for the symptoms of the crises, but in terms of reproducibility. We expand the binary definition of reproducibility into a tier system which allows increasing levels of reproducibility based on external verfiability to counteract the practice of open-washing. We provide solutions for barriers in Computational Social Science that hinder researchers from obtaining the highest level of reproducibility, including the use of alternate data sources and considering reproducibility proactively.

cs.CY↗

arXiv2023

You are a Bot! -- Studying the Development of Bot Accusations on Twitter

The characterization and detection of bots with their presumed ability to manipulate society on social media platforms have been subject to many research endeavors over the last decade. In the absence of ground truth data (i.e., accounts that are labeled as bots by experts or self-declare their automated nature), researchers interested in the characterization and detection of bots may want to tap into the wisdom of the crowd. But how many people need to accuse another user as a bot before we can assume that the account is most likely automated? And more importantly, are bot accusations on social media at all a valid signal for the detection of bots? Our research presents the first large-scale study of bot accusations on Twitter and shows how the term bot became an instrument of dehumanization in social media conversations since it is predominantly used to deny the humanness of conversation partners. Consequently, bot accusations on social media should not be naively used as a signal to train or test bot detection models.

cs.SI↗

arXiv2022

The Hipster Paradox in Electronic Dance Music: How Musicians Trade Mainstream Success off against Alternative Status

The hipster paradox in Electronic Dance Music is the phenomenon that commercial success is collectively considered illegitimate while serious and aspiring professional musicians strive for it. We study this behavioral dilemma using digital traces of performing live and releasing music as they are stored in the \textit{Resident Advisor}, \textit{Juno Download}, and \textit{Discogs} databases from 2001-2018. We construct network snapshots following a formal sociological approach based on bipartite networks, and we use network positions to explain success in regression models of artistic careers. We find evidence for a structural trade-off among success and autonomy. Musicians in EDM embed into exclusive performance-based communities for autonomy but, in earlier career stages, seek the mainstream for commercial success. Our approach highlights how Computational Social Science can benefit from a close connection of data analysis and theory.

cs.SI↗

arXiv2022

Link recommendations: Their impact on network structure and minorities

Network-based people recommendation algorithms are widely employed on the Web to suggest new connections in social media or professional platforms. While such recommendations bring people together, the feedback loop between the algorithms and the changes in network structure may exacerbate social biases. These biases include rich-get-richer effects, filter bubbles, and polarization. However, social networks are diverse complex systems and recommendations may affect them differently, depending on their structural properties. In this work, we explore five people recommendation algorithms by systematically applying them over time to different synthetic networks. In particular, we measure to what extent these recommendations change the structure of bi-populated networks and show how these changes affect the minority group. Our systematic experimentation helps to better understand when link recommendation algorithms are beneficial or harmful to minority groups in social networks. In particular, our findings suggest that, while all algorithms tend to close triangles and increase cohesion, all algorithms except Node2Vec are prone to favor and suggest nodes with high in-degree. Furthermore, we found that, especially when both classes are heterophilic, recommendation algorithms can reduce the visibility of minorities.

cs.SI↗

arXiv2022

Counterfactually Augmented Data and Unintended Bias: The Case of Sexism and Hate Speech Detection

Counterfactually Augmented Data (CAD) aims to improve out-of-domain generalizability, an indicator of model robustness. The improvement is credited with promoting core features of the construct over spurious artifacts that happen to correlate with it. Yet, over-relying on core features may lead to unintended model bias. Especially, construct-driven CAD -- perturbations of core features -- may induce models to ignore the context in which core features are used. Here, we test models for sexism and hate speech detection on challenging data: non-hateful and non-sexist usage of identity and gendered terms. In these hard cases, models trained on CAD, especially construct-driven CAD, show higher false-positive rates than models trained on the original, unperturbed data. Using a diverse set of CAD -- construct-driven and construct-agnostic -- reduces such unintended bias.

cs.CL↗