arXiv subjects

Hamidreza Saffari

Publications and source records attributed to Hamidreza Saffari.

7 recordsLinked to original sources

arXiv2026

Auditing Exposure to Harmful Content on TikTok using Multimodal Language Models: A Cross-National, Age-Stratified Study

Online video platforms can expose young users to harmful content, but independent audits remain difficult because video annotation is costly and moderation judgments vary across languages. We audit TikTok in France, Italy, and Sweden with sockpuppet accounts representing four age personas (13, 16, 19, 40), collecting 36,971 videos from passive For-You-page scrolling and active sessions that scroll, search for harm keywords, and scroll again. To scale annotation, we validate four multimodal LLMs against native-speaker labels on a 300-video reference set. Gemini 2.5 Flash with eight sampled frames plus text performs best (aggregate kappa = 0.42), at half the per-call cost of native-video upload, and we apply it to a 10% sample for approximately \$50 in total API spend across both modalities. Keyword search returns 35-56% harmful content, a 1.5-7.5x increase over the scrolling baseline in ten of twelve country-age combinations; the spike is temporary and flattens the age differences observed in France and Sweden. Under passive scrolling, Italy has the highest harm rate at every age, with Italian age-19 reaching 48.6%. Overall, MLLM-based auditing offers a scalable approach for cross-national youth-safety audits, while provider safety filters (1.1% refusal rate) under-count the most explicit harms.

cs.CL↗

arXiv2025

Beyond Hate Speech: NLP's Challenges and Opportunities in Uncovering Dehumanizing Language

Dehumanization, i.e., denying human qualities to individuals or groups, is a particularly harmful form of hate speech that can normalize violence against marginalized communities. Despite advances in NLP for detecting general hate speech, approaches to identifying dehumanizing language remain limited due to scarce annotated data and the subtle nature of such expressions. In this work, we systematically evaluate four state-of-the-art large language models (LLMs) - Claude, GPT, Mistral, and Qwen - for dehumanization detection. Our results show that only one model-Claude-achieves strong performance (over 80% F1) under an optimized configuration, while others, despite their capabilities, perform only moderately. Performance drops further when distinguishing dehumanization from related hate types such as derogation. We also identify systematic disparities across target groups: models tend to over-predict dehumanization for some identities (e.g., Gay men), while under-identifying it for others (e.g., Refugees). These findings motivate the need for systematic, group-level evaluation when applying pretrained language models to dehumanization detection tasks.

cs.CL↗

arXiv2025

Not All Jokes Land: Evaluating Large Language Models Understanding of Workplace Humor

With the recent advances in Artificial Intelligence (AI) and Large Language Models (LLMs), the automation of daily tasks, like automatic writing, is getting more and more attention. Hence, efforts have focused on aligning LLMs with human values, yet humor, particularly professional industrial humor used in workplaces, has been largely neglected. To address this, we develop a dataset of professional humor statements along with features that determine the appropriateness of each statement. Our evaluation of five LLMs shows that LLMs often struggle to judge the appropriateness of humor accurately.

cs.CL↗

arXiv2025

MultiHoax: A Dataset of Multi-hop False-Premise Questions

As Large Language Models are increasingly deployed in high-stakes domains, their ability to detect false assumptions and reason critically is crucial for ensuring reliable outputs. False-premise questions (FPQs) serve as an important evaluation method by exposing cases where flawed assumptions lead to incorrect responses. While existing benchmarks focus on single-hop FPQs, real-world reasoning often requires multi-hop inference, where models must verify consistency across multiple reasoning steps rather than relying on surface-level cues. To address this gap, we introduce MultiHoax, a benchmark for evaluating LLMs' ability to handle false premises in complex, multi-step reasoning tasks. Our dataset spans seven countries and ten diverse knowledge categories, using Wikipedia as the primary knowledge source to enable factual reasoning across regions. Experiments reveal that state-of-the-art LLMs struggle to detect false premises across different countries, knowledge categories, and multi-hop reasoning types, highlighting the need for improved false premise detection and more robust multi-hop reasoning capabilities in LLMs.

cs.CL↗

arXiv2025

More or Less Wrong: A Benchmark for Directional Bias in LLM Comparative Reasoning

Large language models (LLMs) are known to be sensitive to input phrasing, but the mechanisms by which semantic cues shape reasoning remain poorly understood. We investigate this phenomenon in the context of comparative math problems with objective ground truth, revealing a consistent and directional framing bias: logically equivalent questions containing the words ``more'', ``less'', or ``equal'' systematically steer predictions in the direction of the framing term. To study this effect, we introduce MathComp, a controlled benchmark of 300 comparison scenarios, each evaluated under 14 prompt variants across three LLM families. We find that model errors frequently reflect linguistic steering, systematic shifts toward the comparative term present in the prompt. Chain-of-thought prompting reduces these biases, but its effectiveness varies: free-form reasoning is more robust, while structured formats may preserve or reintroduce directional drift. Finally, we show that including demographic identity terms (e.g., ``a woman'', ``a Black person'') in input scenarios amplifies directional drift, despite identical underlying quantities, highlighting the interplay between semantic framing and social referents. These findings expose critical blind spots in standard evaluation and motivate framing-aware benchmarks for diagnosing reasoning robustness and fairness in LLMs.

cs.CL↗

arXiv2025

Can I introduce my boyfriend to my grandmother? Evaluating Large Language Models Capabilities on Iranian Social Norm Classification

Creating globally inclusive AI systems demands datasets reflecting diverse social norms. Iran, with its unique cultural blend, offers an ideal case study, with Farsi adding linguistic complexity. In this work, we introduce the Iranian Social Norms (ISN) dataset, a novel collection of 1,699 Iranian social norms, including environments, demographic features, and scope annotation, alongside English translations. Our evaluation of 6 Large Language Models (LLMs) in classifying Iranian social norms, using a variety of prompts, uncovered critical insights into the impact of geographic and linguistic context. Results revealed a substantial performance gap in LLMs' comprehension of Iranian norms. Notably, while the geographic context in English prompts enhanced the performance, this effect was absent in Farsi, pointing to nuanced linguistic challenges. Particularly, performance was significantly worse for Iran-specific norms, emphasizing the importance of culturally tailored datasets. As the first Farsi dataset for social norm classification, ISN will facilitate crucial cross-cultural analyses, shedding light on how values differ across contexts and cultures.

cs.SI↗

arXiv2021

Parkinson's Disease Diagnosis based on Gait Cycle Analysis Through an Interpretable Interval Type-2 Neuro-Fuzzy System

In this paper, an interpretable classifier using an interval type-2 fuzzy neural network for detecting patients suffering from Parkinson's Disease (PD) based on analyzing the gait cycle is presented. The proposed method utilizes clinical features extracted from the vertical Ground Reaction Force (vGRF), measured by 16 wearable sensors placed in the soles of subjects' shoes and learns interpretable fuzzy rules. Therefore, experts can verify the decision made by the proposed method based on investigating the firing strength of interpretable fuzzy rules. Moreover, experts can utilize the extracted fuzzy rules for patient diagnosing or adjust them based on their knowledge. To improve the robustness of the proposed method against uncertainty and noisy sensor measurements, Interval Type-2 Fuzzy Logic is applied. To learn fuzzy rules, two paradigms are proposed: 1- A batch learning approach based on clustering available samples is applied to extract initial fuzzy rules, 2- A complementary online learning is proposed to improve the rule base encountering new labeled samples. The performance of the method is evaluated for classifying patients and healthy subjects in different conditions including the presence of noise or observing new instances. Moreover, the performance of the model is compared to some previous supervised and unsupervised machine learning approaches. The final Accuracy, Precision, Recall, and F1 Score of the proposed method are 88.74%, 89.41%, 95.10%, and 92.16%. Finally, the extracted fuzzy sets for each feature are reported.

cs.LG↗