HeRE: A Novel Benchmark for Hebrew Relation Extraction via LLM-Guided Denoising of Knowledge Graph Alignments

Researchers released HeRE, a 1.5-million-instance silver dataset for Hebrew relation extraction, built with a language-agnostic pipeline that denoises noisy knowledge-base/text alignments by treating relation triplets as natural language inference tasks. In the arXiv paper (2610.22542v1), the best open-weights model, Gemma-4-31B-it, reached F1 0.915, matching the best API model, GPT-5.4, at 0.921, and the optimized pipeline scored F1 0.844 on a fully held-out, dual-annotated test set. The authors report multilingual models outperform language-specific ones in few-shot LLM settings, while Hebrew-specific encoders stay competitive under NLI fine-tuning and are strongest in label-free cross-training, and that native-language prompting matches or slightly exceeds English prompts.

arXiv:2610.22542v1 Announce Type: new Abstract: While LLMs demonstrate strong capabilities in English Relation Extraction RE tasks, a significant gap remains in assessing their RE performance in low-resource languages, due to a lack of standardized benchmarks. Creating native, language-specific annotated RE benchmarks is expensive and requires human expertise, while relying on translation yields suboptimal, inaccurate datasets that introduce English-centric biases. To tackle this, we propose a scalable, language-agnostic methodology to create RE data by denoising naturally occurring yet noisy alignments of knowledge bases and corresponding texts. We define a proper alignment via a Natural Language Inference NLI task, where a relation triplet is entailed by an input text. We empirically assess five denoising strategies, and demonstrate that multilingual models outperform language-specific models in the few-shot LLM setting, while Hebrew-specific encoders remain competitive under NLI fine-tuning and are strongest in the label-free cross-training setting. We further find that prompting in the native language achieves performance on par with or slightly exceeding English prompts, confirming that the pipeline operates naturally in the target language. The best open-weights model, Gemma-4-31B-it F1\,=\,0.915 , matches the best API model, GPT-5.4 0.921 , and reaches F1\,=\,0.844 on a fully held-out, dual-annotated test set. Applying the optimized pipeline, we release HeRE: a novel, massive, 1.5-million-instance silver dataset for Hebrew RE, designed to support data-intensive experimentation and model fine-tuning. HeRE bridges a critical gap in standard Hebrew RE evaluation, while providing the NLP community with a reproducible, language-agnostic pipeline for RE dataset construction in other low-resource settings.