Summary of The Power Of Llm-generated Synthetic Data For Stance Detection in Online Political Discussions, by Stefan Sylvius Wagner and Maike Behrendt and Marc Ziegele and Stefan Harmeling
The Power of LLM-Generated Synthetic Data for Stance Detection in Online Political Discussions
by Stefan Sylvius Wagner, Maike Behrendt, Marc Ziegele, Stefan Harmeling
First submitted to arxiv on: 18 Jun 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG)
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 This paper addresses the challenges of stance detection in online political discussions by leveraging large language models (LLMs) to generate synthetic data for training reliable traditional stance detection models. The authors show that fine-tuning these models with LLM-generated synthetic data can improve performance while maintaining interpretability and alignment with real-world data. Specifically, they prompt a Mistral-7B model to generate synthetic data for specific debate questions and demonstrate the effectiveness of this approach in improving stance detection performance. Additionally, they identify the most informative samples in an unlabelled dataset and fine-tune the model with both synthetic data and these samples, achieving better results than a baseline model that is fine-tuned on all true labels. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Stance detection can help improve online political discussions by identifying biased or misleading content. This paper explores how large language models (LLMs) can be used to generate synthetic data for training more reliable stance detection models. The authors show that using LLM-generated synthetic data can improve the performance of traditional stance detection models, making it easier to identify biased or misleading content online. |
Keywords
» Artificial intelligence » Alignment » Fine tuning » Prompt » Synthetic data
