HREB-CRF: Hierarchical Reduced-bias EMA for Chinese Named Entity Recognition

2025-03-03

AI Summary

To address boundary ambiguity, complex semantic representation, and character-form–phoneme–meaning discrepancies that cause recognition errors in Chinese Named Entity Recognition (CNER), this paper proposes a Hierarchical Residual Exponential Bias (HREB) framework integrated with Conditional Random Fields (CRF). We innovatively design a hierarchical fixed-bias weighted averaging mechanism that fuses local and global attention gradients, substantially enhancing word-boundary modeling. Furthermore, the framework incorporates hierarchical attention, Exponential Moving Average (EMA)-based optimization, and gradient aggregation for long texts. Experimental results on MSRA, Resume, and Weibo datasets demonstrate absolute F1-score improvements of 1.1%, 1.6%, and 9.8%, respectively, validating the method’s superior accuracy and robustness in challenging CNER scenarios.

Technology Category

• Natural Language Processing: Information Extraction
• Knowledge Representation and Reasoning: Common-Sense Reasoning
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Representation, semantic annotation, enhancement, enrichments, access and/or integration of a variety of data on the Web

Abstract

Incorrect boundary division, complex semantic representation, and differences in pronunciation and meaning often lead to errors in Chinese Named Entity Recognition(CNER). To address these issues, this paper proposes HREB-CRF framework: Hierarchical Reduced-bias EMA with CRF. The proposed method amplifies word boundaries and pools long text gradients through exponentially fixed-bias weighted average of local and global hierarchical attention. Experimental results on the MSRA, Resume, and Weibo datasets show excellent in F1, outperforming the baseline model by 1.1%, 1.6%, and 9.8%. The significant improvement in F1 shows evidences of strong effectiveness and robustness of approach in CNER tasks.

Problem

• Addresses incorrect boundary division in CNER
• Handles complex semantic representation challenges
• Reduces errors from pronunciation and meaning differences

Innovation

• Hierarchical Reduced-bias EMA for CNER
• Exponentially fixed-bias weighted average
• Amplifies word boundaries and pools gradients