Labeling Training Data for Entity Matching Using Large Language Models

Name: Labeling Training Data for Entity Matching Using Large Language Models
Uploaded: 2026-10-10T07:48:31.200717+00:00
Duration: 560 s
Description: The paper investigates using large language models (LLMs) as teachers to automatically label training data for entity matching tasks. By employing a knowledge distillation workflow, a powerful LLM labels pairs, which are then used to train smaller, cheaper student models. Results show these machine-labeled sets perform nearly as well as those trained on manually curated benchmark data.

The paper investigates using large language models (LLMs) as teachers to automatically label training data for entity matching tasks. By employing a knowledge distillation workflow, a powerful LLM labels pairs, which are then used to train smaller, cheaper student models. Results show these machine-labeled sets perform nearly as well as those trained on manually curated benchmark data.

Open the full episode