Controllable Data Augmentation for Few-Shot Text Mining with Chain-of-Thought Attribute Manipulation
This addresses the challenge of limited training data for few-shot NLP tasks, offering a novel augmentation technique that is incremental but shows strong gains.
The paper tackles the problem of data augmentation for few-shot text mining by proposing Chain-of-Thought Attribute Manipulation (CoTAM), which generates new data by tweaking task-specific attributes like sentiment polarity, resulting in superior performance over other LLM-based methods across various tasks such as text classification and aspect-based sentiment analysis.
Prompting large language models (LLMs) for data augmentation has recently become a common practice in few-shot NLP tasks. In this paper, we propose Chain-of-Thought Attribute Manipulation (CoTAM), a novel approach that generates new data from existing examples by only tweaking in the user-provided, task-specific attribute, e.g., sentiment polarity or topic in movie reviews. Instead of conventional latent representation controlling, we leverage the chain-of-thought prompting to directly edit the text in three steps, (1) attribute decomposition, (2) manipulation proposal, and (3) sentence reconstruction. Extensive results on various tasks, such as text (pair) classification, aspect-based sentiment analysis, and conditional text generation, verify the superiority of CoTAM over other LLM-based augmentation methods with the same number of training examples for both fine-tuning and in-context learning. Remarkably, the 2D visualization of the augmented dataset using principal component analysis revealed a human-recognizable decision boundary that is likely hinted by the attribute manipulation, demonstrating the potential of our proposed approach.