Summary of Leveraging Weakly Annotated Data For Hate Speech Detection in Code-mixed Hinglish: a Feasibility-driven Transfer Learning Approach with Large Language Models, by Sargam Yadav (1) et al.
Leveraging Weakly Annotated Data for Hate Speech Detection in Code-Mixed Hinglish: A Feasibility-Driven Transfer Learning Approach with Large Language Models
by Sargam Yadav, Abhishek Kaushik, Kevin McDaid
First submitted to arxiv on: 4 Mar 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces a novel approach to hate speech detection in low-resource languages, leveraging zero-shot learning with Large Language Models (LLMs). The authors compile a dataset of 100 YouTube comments and weakly label them for coarse and fine-grained misogyny classification in Hinglish. They then apply various approaches, including zero-shot learning, one-shot learning, few-shot learning, and prompting, using models like BART and ChatGPT-3 to classify the comments. The results show that zero-shot classification with BART and few-shot prompting with ChatGPT-3 achieve the best performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper uses Large Language Models (LLMs) to detect hate speech in low-resource languages. It makes a special dataset for this task, using YouTube comments. The model helps us find out if some words are meant to be mean or not. They used different ways to train the model, like not needing any data at all, just one piece of information, and a little bit of training. The best way was when they didn’t need any labeled data at all and used a big language model. |
Keywords
» Artificial intelligence » Classification » Few shot » Language model » One shot » Prompting » Zero shot
