Improving Chinese Character Representation with Formation Tree

TL;DR

This paper introduces Formation Tree-CLIP, a novel model that uses formation trees and a dedicated encoder to improve Chinese character recognition, especially for unseen characters, by leveraging inherent tree structures and efficient training techniques.

Contribution

The paper proposes a formation tree-based representation and encoder for Chinese characters, significantly improving recognition accuracy and training efficiency over previous radical-based sequence methods.

Findings

Enhanced recognition of unseen characters.

Training speed increased by over 2 times.

Better alignment with character inherent properties.

Abstract

Learning effective representations for Chinese characters presents unique challenges, primarily due to the vast number of characters and their continuous growth, which requires models to handle an expanding category space. Additionally, the inherent sparsity of character usage complicates the generalization of learned representations. Prior research has explored radical-based sequences to overcome these issues, achieving progress in recognizing unseen characters. However, these approaches fail to fully exploit the inherent tree structure of such sequences. To address these limitations and leverage established data properties, we propose Formation Tree-CLIP (FT-CLIP). This model utilizes formation trees to represent characters and incorporates a dedicated tree encoder, significantly improving performance in both seen and unseen character recognition tasks. We further introduce masking for to both character images and tree nodes, enabling efficient and effective training. This approach accelerates training significantly (by a factor of 2 or more) while enhancing accuracy. Extensive experiments show that processing characters through formation trees aligns better with their inherent properties than direct sequential methods, significantly enhancing the generality and usability of the representations.