Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LaiDA: Linguistics-aware In-context Learning with Data Augmentation for Metaphor Components Identification

Hongde Liu, Chenyuan He, Feiyang Meng, Changyong Niu, Yuxiang Jia

TL;DR

Metaphor Components Identification (MCI) requires robust contextual understanding, which LaiDA addresses by fusing linguistics-aware in-context learning with data augmentation. The framework bootstraps a high-quality dataset with ChatGPT, pre-trains on a simile corpus to capture basic metaphor patterns, and uses a Graph Attention Network (GAT) encoder to extract linguistic features and retrieve linguistically similar examples via FAISS to guide fine-tuning of a smaller LLM with linguistically informed prompts. Empirical results on NLPCC2024 Shared Task 9 Subtask 2 show LaiDA achieving 2nd place and outperforming several baselines, with ablations confirming the value of both data augmentation and in-context learning. This approach offers an efficient data-collection workflow and a linguistically guided retrieval mechanism that can enhance MCI performance for downstream NLP tasks, with code and data available at the project repository.

Abstract

Metaphor Components Identification (MCI) contributes to enhancing machine understanding of metaphors, thereby advancing downstream natural language processing tasks. However, the complexity, diversity, and dependency on context and background knowledge pose significant challenges for MCI. Large language models (LLMs) offer new avenues for accurate comprehension of complex natural language texts due to their strong semantic analysis and extensive commonsense knowledge. In this research, a new LLM-based framework is proposed, named Linguistics-aware In-context Learning with Data Augmentation (LaiDA). Specifically, ChatGPT and supervised fine-tuning are utilized to tailor a high-quality dataset. LaiDA incorporates a simile dataset for pre-training. A graph attention network encoder generates linguistically rich feature representations to retrieve similar examples. Subsequently, LLM is fine-tuned with prompts that integrate linguistically similar examples. LaiDA ranked 2nd in Subtask 2 of NLPCC2024 Shared Task 9, demonstrating its effectiveness. Code and data are available at https://github.com/WXLJZ/LaiDA.

LaiDA: Linguistics-aware In-context Learning with Data Augmentation for Metaphor Components Identification

TL;DR

Metaphor Components Identification (MCI) requires robust contextual understanding, which LaiDA addresses by fusing linguistics-aware in-context learning with data augmentation. The framework bootstraps a high-quality dataset with ChatGPT, pre-trains on a simile corpus to capture basic metaphor patterns, and uses a Graph Attention Network (GAT) encoder to extract linguistic features and retrieve linguistically similar examples via FAISS to guide fine-tuning of a smaller LLM with linguistically informed prompts. Empirical results on NLPCC2024 Shared Task 9 Subtask 2 show LaiDA achieving 2nd place and outperforming several baselines, with ablations confirming the value of both data augmentation and in-context learning. This approach offers an efficient data-collection workflow and a linguistically guided retrieval mechanism that can enhance MCI performance for downstream NLP tasks, with code and data available at the project repository.

Abstract

Metaphor Components Identification (MCI) contributes to enhancing machine understanding of metaphors, thereby advancing downstream natural language processing tasks. However, the complexity, diversity, and dependency on context and background knowledge pose significant challenges for MCI. Large language models (LLMs) offer new avenues for accurate comprehension of complex natural language texts due to their strong semantic analysis and extensive commonsense knowledge. In this research, a new LLM-based framework is proposed, named Linguistics-aware In-context Learning with Data Augmentation (LaiDA). Specifically, ChatGPT and supervised fine-tuning are utilized to tailor a high-quality dataset. LaiDA incorporates a simile dataset for pre-training. A graph attention network encoder generates linguistically rich feature representations to retrieve similar examples. Subsequently, LLM is fine-tuned with prompts that integrate linguistically similar examples. LaiDA ranked 2nd in Subtask 2 of NLPCC2024 Shared Task 9, demonstrating its effectiveness. Code and data are available at https://github.com/WXLJZ/LaiDA.
Paper Structure (18 sections, 4 equations, 3 figures, 4 tables)

This paper contains 18 sections, 4 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Task Definition
  5. Data Preprocessing
  6. Data Augmentation Pre-training
  7. Graph Attention Network Encoder
  8. Linguistic Feature Matrix Construction
  9. Graph Attention Network
  10. Contrastive Learning
  11. Task Fine-tuning
  12. Experiments
  13. Datasets
  14. Baselines
  15. Implementation Details
  16. ...and 3 more sections

Figures (3)

  • Figure 1: An example of the metaphor components identification.
  • Figure 2: The training workflow of LaiDA.
  • Figure 3: The prompts used during supervised fine-tuning.