Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Stance Classification on Social Media Using Quantified Moral Foundations

Hong Zhang, Quoc-Nam Nguyen, Prasanta Bhattacharya, Wei Gao, Liang Ze Wong, Brandon Siyuan Loh, Joseph J. P. Simons, Jisun An

TL;DR

This study enhances stance detection on social media by incorporating deeper psychological attributes, specifically individuals' moral foundations, to classify stances at both message- and user-levels using both traditional machine learning models and large language models.

Abstract

This study enhances stance detection on social media by incorporating deeper psychological attributes, specifically individuals' moral foundations. These theoretically-derived dimensions aim to provide a comprehensive profile of an individual's moral concerns which, in recent work, has been linked to behaviour in a range of domains, including society, politics, health, and the environment. In this paper, we investigate how moral foundation dimensions can contribute to predicting an individual's stance on a given target. Specifically we incorporate moral foundation features extracted from text, along with message semantic features, to classify stances at both message- and user-levels using both traditional machine learning models and large language models. Our preliminary results suggest that encoding moral foundations can enhance the performance of stance detection tasks and help illuminate the associations between specific moral foundations and online stances on target topics. The results highlight the importance of considering deeper psychological attributes in stance analysis and underscores the role of moral foundations in guiding online social behavior.

Enhancing Stance Classification on Social Media Using Quantified Moral Foundations

TL;DR

This study enhances stance detection on social media by incorporating deeper psychological attributes, specifically individuals' moral foundations, to classify stances at both message- and user-levels using both traditional machine learning models and large language models.

Abstract

This study enhances stance detection on social media by incorporating deeper psychological attributes, specifically individuals' moral foundations. These theoretically-derived dimensions aim to provide a comprehensive profile of an individual's moral concerns which, in recent work, has been linked to behaviour in a range of domains, including society, politics, health, and the environment. In this paper, we investigate how moral foundation dimensions can contribute to predicting an individual's stance on a given target. Specifically we incorporate moral foundation features extracted from text, along with message semantic features, to classify stances at both message- and user-levels using both traditional machine learning models and large language models. Our preliminary results suggest that encoding moral foundations can enhance the performance of stance detection tasks and help illuminate the associations between specific moral foundations and online stances on target topics. The results highlight the importance of considering deeper psychological attributes in stance analysis and underscores the role of moral foundations in guiding online social behavior.
Paper Structure (22 sections, 3 figures, 2 tables)

This paper contains 22 sections, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Stance Detection
  4. Moral Foundation Dictionary (MFD)
  5. Extended MFD (eMFD)
  6. FrameAxis
  7. Datasets
  8. Our Method
  9. Feature Encoding
  10. Stance Detection Models
  11. Prompting LLMs with Moral Features
  12. Generating Descriptive Moral Features
  13. Prompting for Stance Detection
  14. Experiments and Results
  15. Experimental Setup and Data Sampling
  16. ...and 7 more sections

Figures (3)

  • Figure 1: Our method for enhancing stance classification with quantified moral foundations based on traditional ML models and LLMs. Components in red were only used for user-level stance detection, where tweets posted by same user were concatenated before passing through TF-IDF vectorizer, and pooling was applied for moral and contextual embeddings.
  • Figure 2: Target- and stance-level heterogeneity in FrameAxis bias of moral foundations from the SemEval 2016 Task 6 dataset.
  • Figure 3: Biserial correlations between moral foundation features and stance. Correlations that are statistically significant at $p<0.05$ significance level are indicated in bold.