Statistically Supported LLM Ingredient and Recipe Data Collection in Computational Nutrition

2026-07-25

AI Summary

Existing nutritional databases are commonly plagued by incompleteness and inconsistency, and are designed primarily for human consultation, rendering them inadequate for the precise, structured data demands of computational nutrition. This work proposes an automated method for constructing high-quality ingredient-level nutritional data using large language models (LLMs). By treating LLM outputs as samples from a probability distribution, the approach operationalizes uncertainty and enables self-correction through repeated querying, robust statistical estimation, semantic consistency constraints, validation against nutritional logic invariants, and web-based evidence tracing. Evaluated on a test set of 30 ingredients, the method achieves a 98.4% exact-match accuracy for nutrient labels and reduces the median absolute percentage error in nutrient ratios from 31.9% to 10.1%, at an approximate API cost of one U.S. dollar per ingredient.

Abstract

Computational nutrition needs precise ingredient data, but current databases are incomplete, inconsistent, and built for human reference rather than automated reasoning. LLMs could help fill these gaps, but single-pass outputs are unreliable and can introduce silent errors into downstream computation. We present a quality-controlled LLM pipeline for ingredient data acquisition that combines robust statistical estimation, domain-specific invariant checks, and a web-fetch fallback. An illustrative Heap's Law fit to 233 recipes suggests that unique-ingredient growth is sub-linear and front-loaded: the projected ratio of unique ingredients to recipes falls from 1.74 at 100 recipes to 0.19 at 5,000. For each ingredient attribute, repeated LLM queries are treated as samples from a model-induced answer distribution, and we apply robust point estimators and normalised confidence scores across numerical, Boolean, multiple-choice, open categorical, and optional integer types. An invariant guard layer enforces nutritional and logical self-consistency within each ingredient record. Minor numeric inconsistencies are reconciled via a linear program that minimises worst-case percentage deviation while preserving semantic zeros, and major violations are escalated to web-evidence-grounded repair, then human review only if that fails. On a curated 30-ingredient reference set, the pipeline achieves 98.4% exact match on nutrient flags and cuts median absolute percentage error on nutrient ratios from 31.9% for the median-aggregated baseline to 10.1%, a reduction of 21.8 percentage points, at an API cost of about $1 per ingredient. This frames LLM-assisted database construction as a controlled data-engineering workflow that makes uncertainty operational rather than discarding it.

Problem

• computational nutrition
• ingredient data
• LLM reliability
• database inconsistency
• automated reasoning

Innovation

• LLM-based data curation
• robust statistical estimation
• invariant checking
• computational nutrition
• uncertainty-aware engineering