Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

EDDA: A Encoder-Decoder Data Augmentation Framework for Zero-Shot Stance Detection

Daijun Ding, Li Dong, Zhichao Huang, Guangning Xu, Xu Huang, Bo Liu, Liwen Jing, Bowen Zhang

TL;DR

An encoder-decoder data augmentation (EDDA) framework that increases semantic relevance and syntactic variety in augmented texts while enabling interpretable rationale-based learning.

Abstract

Stance detection aims to determine the attitude expressed in text towards a given target. Zero-shot stance detection (ZSSD) has emerged to classify stances towards unseen targets during inference. Recent data augmentation techniques for ZSSD increase transferable knowledge between targets through text or target augmentation. However, these methods exhibit limitations. Target augmentation lacks logical connections between generated targets and source text, while text augmentation relies solely on training data, resulting in insufficient generalization. To address these issues, we propose an encoder-decoder data augmentation (EDDA) framework. The encoder leverages large language models and chain-of-thought prompting to summarize texts into target-specific if-then rationales, establishing logical relationships. The decoder generates new samples based on these expressions using a semantic correlation word replacement strategy to increase syntactic diversity. We also analyze the generated expressions to develop a rationale-enhanced network that fully utilizes the augmented data. Experiments on benchmark datasets demonstrate our approach substantially improves over state-of-the-art ZSSD techniques. The proposed EDDA framework increases semantic relevance and syntactic variety in augmented texts while enabling interpretable rationale-based learning.

EDDA: A Encoder-Decoder Data Augmentation Framework for Zero-Shot Stance Detection

TL;DR

An encoder-decoder data augmentation (EDDA) framework that increases semantic relevance and syntactic variety in augmented texts while enabling interpretable rationale-based learning.

Abstract

Stance detection aims to determine the attitude expressed in text towards a given target. Zero-shot stance detection (ZSSD) has emerged to classify stances towards unseen targets during inference. Recent data augmentation techniques for ZSSD increase transferable knowledge between targets through text or target augmentation. However, these methods exhibit limitations. Target augmentation lacks logical connections between generated targets and source text, while text augmentation relies solely on training data, resulting in insufficient generalization. To address these issues, we propose an encoder-decoder data augmentation (EDDA) framework. The encoder leverages large language models and chain-of-thought prompting to summarize texts into target-specific if-then rationales, establishing logical relationships. The decoder generates new samples based on these expressions using a semantic correlation word replacement strategy to increase syntactic diversity. We also analyze the generated expressions to develop a rationale-enhanced network that fully utilizes the augmented data. Experiments on benchmark datasets demonstrate our approach substantially improves over state-of-the-art ZSSD techniques. The proposed EDDA framework increases semantic relevance and syntactic variety in augmented texts while enabling interpretable rationale-based learning.
Paper Structure (24 sections, 3 equations, 5 figures, 6 tables)

This paper contains 24 sections, 3 equations, 5 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Zero-shot Stance Detection
  4. Text Data Augmentation
  5. Chain-of-Thought model (CoT)
  6. Methodology
  7. Task Description
  8. EDDA
  9. Rationale-Enhanced Network
  10. Experimentation Setup
  11. Experimental Data
  12. Experimental Implementation
  13. Compared Baseline Methods
  14. Experimentatal Results
  15. Result
  16. ...and 9 more sections

Figures (5)

  • Figure 1: The architecture of our proposed encoder-decoder data augmentation (EDDA) framework.
  • Figure 2: An example of the entire EDDA process.
  • Figure 3: Visualization of intermediate vectors in Bert, where red represents the original training set and blue represents the augmented dataset.
  • Figure 4: The experimental results with respect to varying augmentation data size.
  • Figure 5: Comparison of EDDA with data augmentation baselines for previous ZSSD task.