Self-Refinement Strategies for LLM-based Product Attribute Value Extraction

TL;DR

This paper evaluates self-refinement techniques for large language models in extracting product attribute values from unstructured descriptions, finding limited performance gains but increased costs, and highlighting fine-tuning as a more effective approach with sufficient data.

Contribution

It systematically assesses two self-refinement methods for product attribute extraction with LLMs across various learning scenarios, revealing their limited effectiveness and cost implications.

Findings

Self-refinement techniques do not significantly improve extraction performance.

Both techniques substantially increase processing costs.

Fine-tuning with development data achieves the best results.

Abstract

Structured product data, in the form of attribute-value pairs, is essential for e-commerce platforms to support features such as faceted product search and attribute-based product comparison. However, vendors often provide unstructured product descriptions, making attribute value extraction necessary to ensure data consistency and usability. Large language models (LLMs) have demonstrated their potential for product attribute value extraction in few-shot scenarios. Recent research has shown that self-refinement techniques can improve the performance of LLMs on tasks such as code generation and text-to-SQL translation. For other tasks, the application of these techniques has resulted in increased costs due to processing additional tokens, without achieving any improvement in performance. This paper investigates applying two self-refinement techniques (error-based prompt rewriting and self-correction) to the product attribute value extraction task. The self-refinement techniques are evaluated across zero-shot, few-shot in-context learning, and fine-tuning scenarios using GPT-4o. The experiments show that both self-refinement techniques fail to significantly improve the extraction performance while substantially increasing processing costs. For scenarios with development data, fine-tuning yields the highest performance, while the ramp-up costs of fine-tuning are balanced out as the amount of product descriptions increases.