Papers by Tarek Naous
To Lie or Not to Lie? Investigating The Biased Spread of Global Lies by LLMs (2026.acl-long)
Copied to clipboard
Zohaib Khan, Mustafa Dogan, Ifeoma Okoh, Pouya Sadeghi, Siddhartha Shrestha, Sergius Justus Chesami Nyah, Mahmoud O. Mokhiamar, Michael J Ryan, Tarek Naous
| Challenge: | Misinformation is on the rise, and the strong writing capabilities of LLMs lower the barrier for malicious actors to produce and disseminate false information. |
| Approach: | They introduce a multilingual parallel dataset of 440 misinformation generation prompt templates and 6,867 entities, spanning 8 languages and 195 countries. |
| Outcome: | The proposed model reduces misinformation generation across languages and countries . it also reduces the risk of misinformation being spread across countries based on the model's performance . |
Revisiting non-English Text Simplification: A Unified Multilingual Benchmark (2023.acl-long)
Copied to clipboard
| Challenge: | Recent advances in English automatic text simplification have pushed the frontier of multilingual text simulating. |
| Approach: | They propose to use multilingual evaluation benchmarks to evaluate multilingual text simplification models in English and other languages. |
| Outcome: | The proposed benchmark outperforms pre-trained models in Russian in zero-shot cross-lingual transfer to low-resource languages. |
On The Origin of Cultural Biases in Language Models: From Pre-training Data to Linguistic Phenomena (2025.naacl-long)
Copied to clipboard
| Challenge: | Language Models (LMs) have been shown to exhibit a strong preference towards entities associated with Western culture when operating in non-Western languages. |
| Approach: | They propose a parallel Arabic-English benchmark of 58,086 entities associated with Arab and Western cultures and 367 masked natural contexts for entities. |
| Outcome: | The proposed model shows that LMs struggle in Arabic with entities that appear at high frequencies in pre-training, where entities can hold multiple word senses. |
CARE: Multilingual Human Preference Learning for Cultural Awareness (2025.emnlp-main)
Copied to clipboard
| Challenge: | Language Models are tuned with human preferences to produce helpful responses, but the impact of preference tuning on the ability to handle culturally diverse queries remains understudied. |
| Approach: | They introduce a multilingual resource that contains culturally specific questions and 31.7k responses with human judgments. |
| Outcome: | The proposed model outperforms models with stronger initial cultural performance . the proposed model has gaps in the literature on culturally relevant data . |
What are Foundation Models Cooking in the Post-Soviet World? (2025.emnlp-main)
Copied to clipboard
| Challenge: | During the Soviet era, these identities were pressured through forced assimilation under the Russian language and culture. |
| Approach: | They construct a multi-modal dataset encompassing 1147 and 823 dishes in the Russian and Ukrainian languages, centered around the Post-Soviet region. |
| Outcome: | The results show that leading models struggle to correctly identify the origins of dishes from Post-Soviet nations in both text-only and multi-modal Question Answering (QA) the weak correlation between this task and QA suggests that QA alone may be insufficient as an evaluation of cultural understanding. |
ReadMe++: Benchmarking Multilingual Language Models for Multi-Domain Readability Assessment (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing evaluation resources lack domain and language diversity, limiting the ability for cross-domain and cross-lingual analyses. |
| Approach: | They propose to use a multilingual multi-domain dataset to benchmark multilingual and monolingual models for multilingual readability assessment. |
| Outcome: | The proposed model trains better in supervised, unsupervised, and few-shot prompting settings and identifies shortcomings in state-of-the-art unsupervised methods. |
Reducing Privacy Risks in Online Self-Disclosures with Language Models (2024.acl-long)
Copied to clipboard
| Challenge: | Disclosure is a social media activity that can be rewarding but also poses privacy risks. |
| Approach: | They propose to detect and abstract online self-disclosures using a large corpus of 4.8K annotated disclosure spans and a language model to fine-tune for detection. |
| Outcome: | The proposed model can detect and abstract self-disclosures with 80% accuracy, on-par with GPT-3.5. |
Stanceosaurus: Classifying Stance Towards Multicultural Misinformation (2022.emnlp-main)
Copied to clipboard
| Challenge: | Existing corpora focus on misinformation spreading within western countries. |
| Approach: | They present a new corpus of tweets annotated with stance towards 250 misinformation claims. |
| Outcome: | The proposed method achieves 53.1 F1 on Hindi and 50.4 F1 in Arabic without any target-language fine-tuning. |