Adversarial Demonstration Learning for Low-resource NER Using Dual Similarity

2025-07-13

AI Summary

In low-resource named entity recognition (NER) via demonstration learning, two critical challenges persist: significant example selection bias and insufficient model reliance on demonstrations. To address these, we propose a Dual-Similarity Example Selection and Adversarial Demonstration Learning framework. First, we construct high-quality demonstration sets by jointly leveraging semantic similarity (measured via cosine similarity of BERT embeddings) and feature similarity (aligned entity boundary and type distributions). Second, we introduce an adversarial prompt-tuning mechanism that perturbs demonstration inputs to explicitly enforce modeling of example-query associations. Our method integrates dual-similarity retrieval, adversarial training, prompt tuning, and in-context learning. Evaluated on five low-resource NER benchmarks, it substantially outperforms state-of-the-art approaches, achieving an average F1-score improvement of 3.2%. Results demonstrate its effectiveness in enhancing both demonstration utilization efficiency and cross-domain generalization capability.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

We study the problem of named entity recognition (NER) based on demonstration learning in low-resource scenarios. We identify two issues in demonstration construction and model training. Firstly, existing methods for selecting demonstration examples primarily rely on semantic similarity; We show that feature similarity can provide significant performance improvement. Secondly, we show that the NER tagger's ability to reference demonstration examples is generally inadequate. We propose a demonstration and training approach that effectively addresses these issues. For the first issue, we propose to select examples by dual similarity, which comprises both semantic similarity and feature similarity. For the second issue, we propose to train an NER model with adversarial demonstration such that the model is forced to refer to the demonstrations when performing the tagging task. We conduct comprehensive experiments in low-resource NER tasks, and the results demonstrate that our method outperforms a range of methods.

Problem

• Improving low-resource NER via dual similarity demonstration selection
• Enhancing NER tagger's ability to reference demonstration examples
• Addressing demonstration construction and training issues in NER

Innovation

• Uses dual similarity for demonstration selection
• Trains model with adversarial demonstration learning
• Improves NER performance in low-resource scenarios