Beyond Scale and Generation: Understanding Language Model-based Entity Matching

2026-07-27

AI Summary

This study addresses the longstanding ambiguity in entity matching research regarding whether performance gains stem from the matching architecture or from factors such as model backbone, variant, and scale. To disentangle these variables systematically, the authors conduct controlled experiments using the Qwen3 model family across nine datasets, evaluating three matching architectures—bi-encoder, cross-encoder, and generative matcher—combined with three model variants and scales, resulting in 1,215 fine-tuning runs. Findings reveal that cross-encoders consistently achieve the best overall performance, while generative matchers demonstrate superior robustness under distribution shift. Notably, larger models do not necessarily yield better results and often rely on shortcut learning. The work further uncovers the critical influence of model variants on bi-encoder effectiveness and releases all code and data to support reproducibility.

Abstract

Entity matching identifies records that refer to the same real-world entity. Language models can be adapted to this task through bi-encoder, cross-encoder, and generative matcher architectures. However, prior studies often conflate matcher architecture with differences in model backbone, model variant(reflecting different pretraining objectives), and model size, making it difficult to isolate the sources of performance gains. We address this issue through a controlled factorial study spanning three matcher architectures, three model variants and three model sizes from the Qwen3 family, and nine datasets, totaling 1,215 fine-tuning runs. We also evaluate cross-dataset transferability and computational cost. Our results show that model variant is critical for bi-encoders: embedding-oriented variants provide stronger initialization and more favorable representation geometry predictive of downstream matching performance. Cross-encoders retain a consistent advantage over bi-encoders because they jointly encode record pairs rather than representing each record independently, although larger models partially narrow this gap. Generative matchers do not universally outperform cross-encoders. Instead, their advantages concentrate under distribution shift, including subtle unseen differences in record schemas and cross-dataset transfer. We further find that larger models rely more heavily on shortcut learning and therefore do not necessarily perform better. These findings clarify the factors underlying performance differences across matcher architectures and motivate future research and benchmark designs that better disentangle architectural choices from model-level factors while explicitly evaluating distribution shift and cross-dataset transferability. We release our experimental results, code, training scripts, and evaluation data at https://github.com/Jantory/llm-trained-matcher.

Problem

• Entity Matching
• Language Models
• Matcher Architecture
• Model Variant
• Distribution Shift

Innovation

• entity matching
• language models
• matcher architecture
• distribution shift
• controlled factorial study