Reasoning for Hierarchical Text Classification: The Case of Patents

2025-10-08

AI Summary

Patent topic classification is a challenging hierarchical text classification (HTC) task, hindered by deep domain knowledge requirements, an extensive label taxonomy, and limited interpretability. To address these challenges, this work reformulates HTC as a chain-like multi-step reasoning problem and proposes a two-stage large language model (LLM)-based training framework. In Stage I, chain-of-thought prompting and supervised fine-tuning instill hierarchical reasoning capabilities; in Stage II, reinforcement learning optimizes both path consistency and discriminative power across the hierarchy. The method enables sequential label generation with natural-language-level interpretability. Evaluated on patent datasets, it achieves ~3% improvements in accuracy and macro-F1 over strong baselines, and attains state-of-the-art performance across multiple HTC benchmarks—significantly outperforming existing flat or tree-structured approaches.

Technology Category

• Natural Language Processing: (Large) Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Hierarchical text classification (HTC) assigns documents to multiple levels of a pre-defined taxonomy. Automated patent subject classification represents one of the hardest HTC scenarios because of domain knowledge difficulty and a huge number of labels. Prior approaches only output a flat label set, which offers little insight into the reason behind predictions. Therefore, we propose Reasoning for Hierarchical Classification (RHC), a novel framework that reformulates HTC as a step-by-step reasoning task to sequentially deduce hierarchical labels. RHC trains large language models (LLMs) in two stages: a cold-start stage that aligns outputs with chain-of-thought (CoT) reasoning format and a reinforcement learning (RL) stage to enhance multi-step reasoning ability. RHC demonstrates four advantages in our experiments. (1) Effectiveness: RHC surpasses previous baselines and outperforms the supervised fine-tuning counterparts by approximately 3% in accuracy and macro F1. (2) Explainability: RHC produces natural-language justifications before prediction to facilitate human inspection. (3) Scalability: RHC scales favorably with model size with larger gains compared to standard fine-tuning. (4) Applicability: Beyond patents, we further demonstrate that RHC achieves state-of-the-art performance on other widely used HTC benchmarks, which highlights its broad applicability.

Problem

• Automated patent classification faces domain complexity and label abundance
• Existing methods lack hierarchical reasoning for prediction transparency
• Proposed framework reformulates classification as sequential reasoning task

Innovation

• Reformulates HTC as step-by-step reasoning task
• Trains LLMs with cold-start and reinforcement learning stages
• Produces natural-language justifications before predictions