Auto prompting without training labels: An LLM cascade for product quality assessment in e-commerce catalogs

TL;DR

This paper presents a training-free cascade method for automatically generating prompts to evaluate product quality in e-commerce, significantly reducing expert effort and improving accuracy across languages and tasks.

Contribution

It introduces a novel auto-prompting cascade that automatically refines prompts for LLMs without training, tailored for large-scale product quality assessment in e-commerce catalogs.

Findings

Improves precision and recall by 8-10% over chain-of-thought prompting.

Reduces domain expert effort from 5.1 hours to 3 minutes per attribute.

Generalizes effectively across five languages and multiple tasks.

Abstract

We introduce a novel, training free cascade for auto-prompting Large Language Models (LLMs) to assess product quality in e-commerce. Our system requires no training labels or model fine-tuning, instead automatically generating and refining prompts for evaluating attribute quality across tens of thousands of product category-attribute pairs. Starting from a seed of human-crafted prompts, the cascade progressively optimizes instructions to meet catalog-specific requirements. This approach bridges the gap between general language understanding and domain-specific knowledge at scale in complex industrial catalogs. Our extensive empirical evaluations shows the auto-prompt cascade improves precision and recall by $8-10\%$ over traditional chain-of-thought prompting. Notably, it achieves these gains while reducing domain expert effort from 5.1 hours to 3 minutes per attribute - a $99\%$ reduction. Additionally, the cascade generalizes effectively across five languages and multiple quality assessment tasks, consistently maintaining performance gains.