Enabling Stroke-Level Structural Analysis of Hieroglyphic Scripts without Language-Specific Priors

2026-01-09

arXiv.org

AI Summary

Existing large language models and multimodal architectures struggle to effectively model the stroke-level structure of hieroglyphic scripts, while conventional approaches rely heavily on linguistic priors and incur high manual annotation costs. This work proposes HieroSA, a novel framework that, for the first time, automatically extracts stroke-level structural representations directly from character bitmaps without requiring linguistic priors or handcrafted annotations. By leveraging image processing and geometric normalization techniques, HieroSA generates interpretable sequences of line segments in a normalized coordinate space. The method overcomes script-specific limitations, enabling a universal and interpretable modeling approach applicable to both ancient and modern hieroglyphic systems. Experimental results demonstrate its ability to accurately capture intrinsic character structures and their semantic properties, offering a new computational tool for glyphographic analysis and facilitating cross-lingual transfer applications.

Technology Category

• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Machine Learning: Structured Learning
• Cognitive Modeling & Cognitive Systems: Symbolic Representations

Application Category

• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Hieroglyphs, as logographic writing systems, encode rich semantic and cultural information within their internal structural composition. Yet, current advanced Large Language Models (LLMs) and Multimodal LLMs (MLLMs) usually remain structurally blind to this information. LLMs process characters as textual tokens, while MLLMs additionally view them as raw pixel grids. Both fall short to model the underlying logic of character strokes. Furthermore, existing structural analysis methods are often script-specific and labor-intensive. In this paper, we propose Hieroglyphic Stroke Analyzer (HieroSA), a novel and generalizable framework that enables MLLMs to automatically derive stroke-level structures from character bitmaps without handcrafted data. It transforms modern logographic and ancient hieroglyphs character images into explicit, interpretable line-segment representations in a normalized coordinate space, allowing for cross-lingual generalization. Extensive experiments demonstrate that HieroSA effectively captures character-internal structures and semantics, bypassing the need for language-specific priors. Experimental results highlight the potential of our work as a graphematics analysis tool for a deeper understanding of hieroglyphic scripts. View our code at https://github.com/THUNLP-MT/HieroSA.

Problem

• hieroglyphic scripts
• stroke-level structural analysis
• language-specific priors
• logographic writing systems
• character structure modeling

Innovation

• stroke-level analysis
• hieroglyphic scripts
• multimodal LLMs
• structure representation
• cross-lingual generalization