Hate Personified: Investigating the role of LLMs in content moderation

For subjective tasks such as hate detection, where people of different socio-cultural backgrounds perceive hate differently, the Large Language Model's (LLM) ability to represent diverse groups is unclear. By including additional context in prompts, we comprehensively analyze LLM's sensitivity to geographical priming, persona attributes, and numerical information to assess how well the needs of various groups are reflected. Our findings on two LLMs, five languages, and six datasets reveal that mimicking persona-based attributes leads to annotation variability. Meanwhile, incorporating geographical signals leads to better regional alignment. We also find that the LLMs are sensitive to numerical anchors, indicating the ability to leverage community-based flagging efforts and exposure to adversaries. Our work provides preliminary guidelines and highlights the nuances of applying LLMs in culturally sensitive cases.

Repo Overview

• Original_Datasets

  • CREHate(En), HateXplain(En), MLMA(Ar, Fr) & HASOC(De, Hi) dataset files
  • Dataset_stats.ipynb

• Graphs

  • GC, NC & ML
  • GC.ipynb, NC.ipynb & ML.ipynb

• RQ1_GC: Geographical Cues

  • Analysis
    • All_models_performance.ipynb
    • PerformanceOS_LLMs.csv
    • PerformanceChatGPT.csv
  • OS_LLMs
  • ChatGPT

• RQ2_PC: Persona Cues

  • Analysis
    • All_models_performance.ipynb
    • PerformanceOS_LLMs_HA.csv
    • PerformanceOS_LLMs_N.csv
    • PerformanceChatGPT_HA.csv
    • PerformanceChatGPT_N.csv
  • OS_LLMs
  • ChatGPT

• RQ3_NC: Numerical Cues

  • Analysis
    • All_models_performance.ipynb
    • PerformanceOS_LLMs_H.csv
    • PerformanceOS_LLMs_N.csv
    • PerformanceChatGPT_H.csv
    • PerformanceChatGPT_N.csv
  • OS_LLMs
  • ChatGPT
  • Temp_Variations

• RQ_ML: Multi-lingual

  • Analysis
    • All_models_performance.ipynb
    • All_models_performance_SL.ipynb
    • All_models_performance_LocalLLMs.ipynb
    • Performance.csv
    • Performance_SL.csv
    • Performance_LocalLLMs.csv
  • OS_LLMs
  • ChatGPT

• Prompt_Files_Guide.xlsx

• Prompt_files_generator.ipynb

• English_to_same_language_convertor.ipynb

• Inference_OS_LLMs.py

• Inference_ChatGPT.py

• Inference_temp_variation.py

• requirements.txt

• README.md

Cite

@inproceedings{masud-etal-2024-hate,
    title = "Hate Personified: Investigating the role of {LLM}s in content moderation",
    author = "Masud, Sarah  and Singh, Sahajpreet  and Hangya, Viktor  and Fraser, Alexander  and Chakraborty, Tanmoy",
    editor = "Al-Onaizan, Yaser  and Bansal, Mohit  and Chen, Yun-Nung",
    booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing",
    month = nov,
    year = "2024",
    address = "Miami, Florida, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2024.emnlp-main.886",
    pages = "15847--15863"
}