Austria

Algorithmic hate speech detection is widely used today. However, biases within these systems can lead to discrimination. This research presents an overview of bias mitigation strategies in the field of hate speech detection. The identified principles are grouped into four categories, based on their operation principles. A novel taxonomy of bias mitigation methods is proposed. The mitigation strategies are characterized based on their key concepts and analyzed in terms of their application stage and their need for knowledge of protected attributes. Additionally, the paper discusses potential combinations of these strategies. This research shifts the focus from identifying present biases to examining the similarities and differences between mitigation strategies, thereby facilitating the exchange, stacking, and ensembling of these strategies in future research.

ACL 2025

A Comprehensive Taxonomy of Bias Mitigation Methods for Hate Speech Detection

### Welcome to ACL 2023, the 61st Annual Meeting of the Association for Computational Linguistics! 
 The conference will be held in Toronto, Canada, July 9-14, 2023. 
Following the succession of the recent conferences in our field, ACL 2023 will adopt a hybrid format.
While the impact of Covid has considerably diminished in terms of traveling, obtaining visas to Canada
entails a very long process. Moreover, the global economic conditions pose challenges for many individuals to travel to conferences. Recognizing these circumstances, we know many participants may not be
able to attend the conference in person. Therefore, we are committed to providing a great virtual platform
so everyone has the opportunity to interact with other participants and enjoy the conference. Based on the
current registered participants, approxiately 30% have chosen to attend the conference virtually. Whether
you join us in person or virtually, we sincerely hope everyone has a remarkable conference experience. 
This General Chair’s message is where I express my gratitude to the many individuals who have made
enormous contributions to the conference over the past year.

Read [**ACL 2023 General Chair's message**](https://docs.google.com/document/d/1WobYM7norbG4dI48s75HfJoD89qgX5a_F-6U8AteLSA/edit?usp=sharing/) in full.

##### **[Conference Handbook](https://2023.aclweb.org/downloads/acl2023-handbook.pdf)**

ACL 2023

The Association for Computational Linguistics (ACL) is the premier international scientific and professional society for people working on computational problems involving human language, a field often referred to as either computational linguistics or natural language processing.

workshop paper

### Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Message from the General Chair: 
*It is my great pleasure and honor to welcome you to the 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025), held in beautiful Vienna, Austria, from July 27 to August 1, 2025. ACL2025continues our field’s tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.* 
[Read more](https://drive.google.com/file/d/1GI_hvOpjswAuYdUTromfeDiPpCcqidwg/view?usp=sharing)

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to ACL 2025". Please be sure to check your spam email folder if you do not see an email confirmation right away.

Please log in

To access this event page, you need to log in with the **email address you registered with**. 

Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

The detection of sensitive content in large datasets is crucial for ensuring that shared and analysed data is free from harmful material. However, current moderation tools, such as external APIs, suffer from limitations in customisation, accuracy across diverse sensitive categories, and privacy concerns. Additionally, existing datasets and open-source models focus predominantly on toxic language, leaving gaps in detecting other sensitive categories such as substance abuse or self-harm. In this paper, we put forward a unified dataset tailored for social media content moderation across six sensitive categories: conflictual language, profanity,sexually explicit material, drug-related content, self-harm, and spam. By collecting and annotating data with consistent retrieval strategies and guidelines, we address the shortcomings of previous focalised research. Our analysis demonstrates that fine-tuning large language models (LLMs) on this novel dataset yields significant improvements in detection performance compared to open off-the-shelf models such as LLaMA, and even proprietary OpenAI models, which underperform by 10-15% overall. This limitation is even more pronounced on popular moderation APIs, which cannot be easily tailored to specific sensitive content categories, among others.

Sensitive Content Classification in Social Media: A Holistic Resource and Evaluation

Algorithmic content moderation governs online speech on large-scale commercial platforms, often under the guise of neutrality. Yet, it routinely reproduces white, middle-class norms of civility and penalizes marginalized voices for unruly and resistant speech. This paper critiques the prevailing ‘pathological' approach to moderation that prioritizes sanitization over justice. Drawing on Marxist-feminist ethics, this paper advances three theses for the future of context-aware algorithmic moderation: (1) prioritizing participatory parity over civility, (2) incorporating identity- and context-aware analysis of speech; and (3) replacing purely numerical evaluations with justice-oriented, community-sensitive metrics. While acknowledging the structural limitations posed by platform capitalism, this paper positions the proposed framework as both critique and provocation, guiding regulatory reform, civil advocacy, and visions for mission-driven online content moderation serving digital commons.

From civility to parity: Marxist-feminist ethics for context-aware algorithmic

The consistently high prevalence of hate speech on the Internet continues to pose significant social and individual challenges. Given the centrality of social networks in public discourse, automating the identification of criminally relevant content is a pressing challenge. This study addresses the challenge of developing an automated system that is capable of classifying online comments in a criminal justice context and categorising them into relevant sections of the criminal code. Not only technical, but also ethical and legal requirements must be considered. To this end, 351 comments were annotated by public prosecutors from the Central Office for Combating Internet and Computer Crime (ZIT) according to previously formed paragraph classes. These groupings consist of several German criminal law statutes that most hate comments violate. In the subsequent phase of the research, a further 839 records were assigned to the classes by student annotators who had been trained previously.

A Novel Dataset for Classifying German Hate Spe

In-context learning (ICL) has shown significant benefits, particularly in scenarios where large amounts of labeled data are unavailable. However, its effectiveness for highly subjective tasks, such as toxic language detection, remains an open question. A key challenge in this setting is to select shots to maximize performance. Although previous work has focused on enhancing variety and representativeness, the role of annotator disagreement in shot selection has received less attention. In this paper, we conduct an in-depth analysis of ICL using two families of open-source LLMs (Llama-3* and Qwen2.5) of varying sizes, evaluating their performance in five prominent English datasets covering multiple toxic language phenomena. We use disaggregated annotations and categorize different types of training examples to assess their impact on model predictions. We specifically investigate whether selecting shots based on annotators' entropy -- focusing on ambiguous or difficult examples -- can improve generalization in LLMs. Additionally, we examine the extent to which the order of examples in prompts influences model behavior.
Our results show that selecting shots based on entropy from annotator disagreement can enhance ICL performance. Specifically, ambiguous shots with a median entropy value generally lead to the best results for our selected LLMs in the few-shot setting. However, ICL often underperforms when compared to fine-tuned encoders.

Learning from Disagreement: Entropy-Guided Few-Shot Selection for Toxic Language Detection

The curation of hate speech datasets involves complex design decisions that balance competing priorities. This paper critically examines these methodological choices in a diverse range of datasets, highlighting common themes and practices, and their implications for dataset reliability. Drawing on Max Weber's notion of ideal types, we argue for a reflexive approach in dataset creation, urging researchers to acknowledge their own value judgments during dataset construction, fostering transparency and methodological rigour.

Web(er) of Hate: A Survey on How Hate Speech Is Typed

Automated counter-narratives (CN) offer a promising strategy for mitigating online hate speech, yet concerns about their affective tone, accessibility, and ethical risks remain. We propose a framework for evaluating Large Language Model (LLM)-generated CNs across four dimensions: persona framing, verbosity and readability, affective tone, and ethical robustness. Using GPT-4o-Mini, Cohere's CommandR-7B, and Meta's LLaMA 3.1-70B, we assess three prompting strategies on the MT-Conan and HatEval datasets. Our findings reveal that LLM-generated CNs are often verbose and adapted for people with college-level literacy, limiting their accessibility. While emotionally guided prompts yield more empathetic and readable responses, there remain concerns surrounding safety and effectiveness.

Think Like a Person Before Responding: A Multi-Faceted Evaluation of Persona-Guided LLMs for Countering Hate Speech.

The escalating spread of homophobic and transphobic rhetoric in both online and offline spaces has become a growing global concern, with Italy standing out as one of the countries where acts of violence against LGBTQIA+ individuals persist and increase year after year. This short paper study analyzes hateful language against LGBTQIA+ individuals in Italian using novel annotation labels for aggressiveness and target. We assess a range of multilingual and Italian language models on this new
annotation layers across zero-shot, few-shot, and fine-tuning settings. The results reveal significant performance gaps across models and settings, highlighting the limitations of zero- and few-shot approaches and the importance of fine-tuning on labelled data, when available, to achieve high prediction performance.

HODIAT: A New Dataset for Detecting Homotransphobic Hate Speech in Italian with Aggressiveness and Target Annotation

Online communities provide support and help to individuals transitioning gender. However, this point of transition also increases vulnerability, coupled with increased exposure to online harms. In this research, we analyse a popular hate and harassment site known for targeting minority groups, including transgender people. We analyse 17 million posts dating back to 2012 to gain insights into the types of information collected about targets. We find users commonly link to social media sites such as Twitter/X and meticulously archive links related to their targets. We scrape over 150,000 relevant links posted to Twitter/X and their archived versions and analyse the profiles and posts. We find targets often tweet about harassment, popculture, and queer and gender-related discussions. We develop and evaluate classifiers to detect calls for harassment, doxxing, mention of transgender individuals, and toxic/abusive speech within the forum posts. The results of our classifiers show that forum posts about transgender individuals are significantly more likely to contain other harmful content.

Beyond the Binary: Analysing Transphobic Hate and Harassment Online

We introduce a novel class of adversarial attacks on toxicity detection models that exploit language models' failure to interpret spatially structured text in the form of ASCII art. To evaluate the effectiveness of these attacks, we propose ToxASCII, a benchmark designed to assess the robustness of toxicity detection systems against visually obfuscated inputs. Our attacks achieve a perfect Attack Success Rate (ASR) across a diverse set of state-of-the-art large language models and dedicated moderation tools, revealing a significant vulnerability in current text-only moderation systems.

Evading Toxicity Detection with ASCII-art: A Benchmark of Spatial Attacks on Moderation Systems

Counterspeech is a key strategy against harmful online content, but scaling expert-driven efforts is challenging. Large Language Models (LLMs) present a potential solution, though their use in countering conspiracy theories is under-researched. Unlike for hate speech, no datasets exist that pair conspiracy theory comments with expert-crafted counterspeech. We address this gap by evaluating the ability of GPT-4o, Llama 3, and Mistral to effectively apply counterspeech strategies derived from psychological research provided through structured prompts. Our results show that the models often generate generic, repetitive, or superficial results. Additionally, they over-acknowledge fear and frequently hallucinate facts, sources, or figures, making their prompt-based use in practical applications problematic.

Downloads

Next from ACL 2025

Sensitive Content Classification in Social Media: A Holistic Resource and Evaluation

Similar lecture

Robust Hate Speech Detection: A Cross-Dataset Empirical Comparison

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from ACL 2025

Sensitive Content Classification in Social Media: A Holistic Resource and Evaluation

Similar lecture

Robust Hate Speech Detection: A Cross-Dataset Empirical Comparison

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

Downloads