Optimizing Product Provenance Verification using Data Valuation Methods

TL;DR

The paper tackles provenance verification in global supply chains by combining Stable Isotope Ratio Analysis with Gaussian process isoscapes and deploying a Shapley-based data valuation framework to optimize training data. It introduces forward and backward models for geographic origin inference, leverages atmospheric variables, and employs Truncated Monte Carlo Shapley to identify high value data points that improve model robustness and accuracy. The approach is validated on Quercus datasets and deployed within the World Forest ID system, yielding tangible enforcement outcomes and highlighting improvements over naive baselines. The work demonstrates that data valuation not only enhances model performance but also guides cost effective sampling in real world regulatory contexts, with implications for broader natural resource supply chains.

Abstract

Determining and verifying product provenance remains a critical challenge in global supply chains, particularly as geopolitical conflicts and shifting borders create new incentives for misrepresentation of commodities, such as hiding the origin of illegally harvested timber or stolen agricultural products. Stable Isotope Ratio Analysis (SIRA), combined with Gaussian process regression-based isoscapes, has emerged as a powerful tool for geographic origin verification. While these models are now actively deployed in operational settings supporting regulators, certification bodies, and companies, they remain constrained by data scarcity and suboptimal dataset selection. In this work, we introduce a novel deployed data valuation framework designed to enhance the selection and utilization of training data for machine learning models applied in SIRA. By quantifying the marginal utility of individual samples using Shapley values, our method guides strategic, cost-effective, and robust sampling campaigns within active monitoring programs. By prioritizing high-informative samples, our approach improves model robustness and predictive accuracy across diverse datasets and geographies. Our framework has been implemented and validated in a live provenance verification system currently used by enforcement agencies, demonstrating tangible, real-world impact. Through extensive experiments and deployment in a live provenance verification system, we show that this system significantly enhances provenance verification, mitigates fraudulent trade practices, and strengthens regulatory enforcement of global supply chains.

Figures (6)

• Figure 1: Distribution of data Shapley values, green and red lines represent the mean and the zero, respectively (top: backward model, bottom: forward model); see Section \ref{['ssec:0_dist']}.
• Figure 2: Effect of model architecture on data valuation framework ($N=87$); see Section \ref{['ssec:1_gp_rf']}.
• Figure 3: Enhancing missing data strategies through data valuation ($N=491$); see Section \ref{['ssec:2_imputation']}.
• Figure 4: Random Forest-based data valuation on USA only data ($N=87$); see Section \ref{['ssec:3_direction']}.
• Figure 5: Random Forest-based data valuation on Europe only data ($N=287$); see Section \ref{['ssec:3_direction']}.