SC-Taxo: Hierarchical Taxonomy Generation under Semantic Consistency Constraints using Large Language Models
For researchers and practitioners needing high-quality scientific taxonomies to organize and navigate rapidly expanding literature, SC-Taxo provides a more semantically consistent and structurally coherent generation method.
SC-Taxo addresses structural inconsistencies and semantic misalignment in hierarchical taxonomy generation by introducing a bidirectional heading generation mechanism with hierarchy-aware refinement using LLMs, achieving consistent improvements in hierarchy alignment and heading quality across multiple benchmarks, including cross-lingual generalization on Chinese scientific literature.
Scientific literature is expanding at an unprecedented pace, making it increasingly challenging to efficiently organize and access domain knowledge. A high-quality scientific taxonomy offers a structured and hierarchical representation of a research field, facilitating literature exploration and topic navigation, as well as enabling downstream applications such as trend analysis, idea generation, and information retrieval. However, existing taxonomy generation approaches often suffer from structural inconsistencies and semantic misalignment across hierarchical levels. Through empirical analysis, we find that these issues largely stem from inadequate modeling of hierarchical semantic consistency. To address this limitation, we propose a semantic-consistent taxonomy generation (SC-Taxo) framework that leverages large language models (LLMs) with hierarchy-aware refinement stages to ensure semantic consistency. Specifically, SC-Taxo introduces a bidirectional heading generation mechanism that jointly performs bottom-up abstraction and top-down semantic constraint, while further capturing peer-level semantic dependencies to enhance horizontal consistency. Experiments on multiple benchmark datasets demonstrate consistent improvements in hierarchy alignment and heading quality, and additional evaluation on Chinese scientific literature validates its robust cross-lingual generalization.