Can ChatGPT Reproduce Human-Generated Labels? A Study of Social Computing Tasks

04/20/2023
by   Yiming Zhu, et al.
0

The release of ChatGPT has uncovered a range of possibilities whereby large language models (LLMs) can substitute human intelligence. In this paper, we seek to understand whether ChatGPT has the potential to reproduce human-generated label annotations in social computing tasks. Such an achievement could significantly reduce the cost and complexity of social computing research. As such, we use ChatGPT to relabel five seminal datasets covering stance detection (2x), sentiment analysis, hate speech, and bot detection. Our results highlight that ChatGPT does have the potential to handle these data annotation tasks, although a number of challenges remain. ChatGPT obtains an average accuracy 0.609. Performance is highest for the sentiment analysis dataset, with ChatGPT correctly annotating 64.9 show that performance varies substantially across individual labels. We believe this work can open up new lines of analysis and act as a basis for future research into the exploitation of ChatGPT for human annotation tasks.

READ FULL TEXT
research
09/04/2020

Visual Sentiment Analysis from Disaster Images in Social Media

The increasing popularity of social networks and users' tendency towards...
research
05/24/2023

Sentiment Analysis in the Era of Large Language Models: A Reality Check

Sentiment analysis (SA) has been a long-standing research area in natura...
research
05/31/2023

Automated Annotation with Generative AI Requires Validation

Generative large language models (LLMs) can be a powerful tool for augme...
research
11/03/2021

End-to-End Annotator Bias Approximation on Crowdsourced Single-Label Sentiment Analysis

Sentiment analysis is often a crowdsourcing task prone to subjective lab...
research
07/16/2023

SentimentGPT: Exploiting GPT for Advanced Sentiment Analysis and its Departure from Current Machine Learning

This study presents a thorough examination of various Generative Pretrai...
research
05/16/2023

Executive Voiced Laughter and Social Approval: An Explorative Machine Learning Study

We study voiced laughter in executive communication and its effect on so...
research
12/21/2022

What do LLMs Know about Financial Markets? A Case Study on Reddit Market Sentiment Analysis

Market sentiment analysis on social media content requires knowledge of ...

Please sign up or login with your details

Forgot password? Click here to reset