---
title: "Automated merchant review summary system with integrated feedback"
description: "The system uses representative sampling, large language model (LLM) summarization, automated quality checks, and consumer feedback to generate concise summaries from written merchant reviews. Consumers can scan an overall summary or focus on specific topics such as taste, packaging, and portion size, while merchants can identify recurring themes without reading every comment. The pipeline calculates the merchant’s rating distribution, uses a sentiment model to score reviews, and applies stratified sampling to select a representative subset before sending them to an LLM for summary generation. A separate LLM validates that each claim in the generated summary is supported by the source reviews, and summaries that fail this automated assessment are withheld from publication to reduce unsupported claims. The system regenerates summaries when new reviews or negative consumer feedback reach configured thresholds, and the feedback-driven regeneration includes anti-abuse controls to maintain trustworthiness."
---

# Automated merchant review summary system with integrated feedback

[Grab](https://yomu.fyi/company/grab) · Chengju Zhou · Sep 25, 2026

**Type:** Problem & solution

## Summary

The system uses representative sampling, large language model (LLM) summarization, automated quality checks, and consumer feedback to generate concise summaries from written merchant reviews. Consumers can scan an overall summary or focus on specific topics such as taste, packaging, and portion size, while merchants can identify recurring themes without reading every comment. The pipeline calculates the merchant’s rating distribution, uses a sentiment model to score reviews, and applies stratified sampling to select a representative subset before sending them to an LLM for summary generation. A separate LLM validates that each claim in the generated summary is supported by the source reviews, and summaries that fail this automated assessment are withheld from publication to reduce unsupported claims. The system regenerates summaries when new reviews or negative consumer feedback reach configured thresholds, and the feedback-driven regeneration includes anti-abuse controls to maintain trustworthiness.

## Context

Grab, a Southeast Asian superapp offering ride-hailing, food delivery, and digital financial services, built this system to help consumers and merchants extract actionable insights from large volumes of merchant reviews.

## Approach / What changed

The system uses representative sampling via stratified sampling based on rating distribution, LLM-based summary generation and validation, and feedback-driven regeneration triggers.

## Takeaways

- The system uses stratified sampling based on rating distribution to select a representative subset of reviews, reducing the chance that one rating group dominates the summary.
- A separate LLM validates that each claim in the generated summary is supported by the source reviews; summaries that fail are withheld from publication.
- The system regenerates summaries when new reviews or negative consumer feedback reach configured thresholds, with anti-abuse controls needed for trustworthiness.

## FAQ

### How does the system ensure review summaries are representative?

It calculates the merchant's rating distribution, uses a sentiment model to score reviews, and applies stratified sampling to select a subset that follows that distribution, reducing dominance by one rating group.

### What happens if a generated summary contains unsupported claims?

A separate LLM validates each claim against the source reviews; summaries that fail the automated assessment are withheld from publication.

### How does the system keep summaries up to date?

It monitors new reviews and consumer feedback; when either reaches a configured threshold, it triggers regeneration of the summary.

**Tags:** [AI](https://yomu.fyi/topic/ai), [Data Engineering](https://yomu.fyi/topic/data-engineering), [LLMs](https://yomu.fyi/topic/llm)

- Source: [Grab](https://engineering.grab.com/automated-merchant-review-summary-system-with-integrated-feedback)
- Source URL: https://engineering.grab.com/automated-merchant-review-summary-system-with-integrated-feedback
- Ingested by Yomu: 2026-10-05T06:00:08.190Z

[Read original post](https://engineering.grab.com/automated-merchant-review-summary-system-with-integrated-feedback)