HeRE: A Novel Benchmark for Hebrew Relation Extraction via LLM-Guided Denoising of Knowledge Graph Alignments
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.
ProofPaper ↗
Key points
- 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.
- 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.
- 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.
Sources (1)
- [1]HeRE: A Novel Benchmark for Hebrew Relation Extraction via LLM-Guided Denoising of Knowledge Graph AlignmentsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 15, 06:40 PM
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.
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.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 15, 2026Towards Continual Learning: Mitigating Forgetting With Subspace Tuning
- Oct 15, 2026Name-Bias in Large Language Models
- Oct 15, 2026NVIDIA Nemotron 3.5 Super VL Powers Specialized Agents with Multimodal Reasoning
- Oct 15, 2026Incentivizing Tool-Use Reinforcement Learning for Post-Arrival Emergency Care Prediction
- Oct 14, 2026Optimal Hydra: Model Heterogeneous Federated Learning for LLMs using Data-Informed Hydra-Nets
- Oct 14, 2026mistralai/Timestral-209M
