Spatial-temporal data authenticity analytics for apparel ESG disclosure

Abstract

Environmental, Social, and Governance (ESG) disclosure has received significant attention from both the academia and practitioners due to a close connection with sustainable investment and development. Against the backdrop of the contemporary era characterized by the overarching pursuit of sustainable development, the apparel industry has captivated the focused attention of all segments of society. The practice of ESG disclosure within the apparel industry is a strategic imperative that intersects the twin pillars of economic development and environmental protection. However, the data authenticity of ESG disclosure continues to be a challenge when generating the report since the existence of greenwashing, specifically manifested through risks to data authenticity, consistency, and transparency. To address this, this study proposes an Event-based Spatial-temporal Analytics (ESTA) for data authenticity of ESG disclosure in Apparel Industry by utilizing visibility and traceability in cyber-physical system (CPS) by the following four studies. Following an in-depth investigation of ESG disclosure paradigms, Study I proposes an IoT and Blockchain-enabled ESG Disclosure System (IBESG). IBESG leverage embedded sensors and edge computing to automate ESG data collection and preprocessing, ensuring standardized measurement. Second, blockchain-based immutable ledger mechanisms are deployed to create cryptographic chaining between raw operational data and final disclosures. The IBESG is designed as a four-layer CPS architecture to meet the requirements in ESG reporting scenarios. Study II develops a Local-Global Authenticity Verification (LGA) framework that implements a dual-layer authentication mechanism through edge-cloud integration based on the IBESG. This framework employs event-driven modeling for qualitative integrity analysis of data flows while establishing quantitative validation through systematic computational paradigms. Three event modeling methodologies based on LGA are adopted, including rule-based spatial-temporal event synthesis, Petri net-based event generation, and FL-based spatial-temporal clustering. Leveraging the advantages of the IBESG and LGA, Study III constructs an Event-based Spatial-Temporal Data Authenticity Analytics (ESTA) algorithm to verify the authenticity of raw data in the ESG disclosure preparation phase. The ESTA algorithm incorporates three distinct processing pathways: rule-based spatial-temporal analytics, Petri net-based spatial-temporal analytics, and federated semi-supervised learning-based analytics. The spatial-temporal analytics serves as an integrative thread permeating all three technical pathways, establishing the core mechanism for event-driven data authentication. The details of each analytics algorithm are mathematically presented. Three numerical studies are conducted, including the control-variable main experiments and sensitivity analyses, to illustrate the performance of each algorithm, achieving AUC from 0.87 to 0.97. The IBESG-LGA-ESTA solution is empirically implemented in Study IV through a real-world case study within the apparel manufacturing industry, demonstrating its practical applicability and effectiveness. Empirical validation in Hanbo shows 25–50% efficiency improvement in ESG reporting. This research is significant both in practical and in academia fields. From the aspect of practicality, the authenticity of ESG information disclosure is crucial for ensuring a fair investment market. From an academic perspective, research on data authenticity can be applied not only to ESG information disclosure but also can provide a means of determining the authenticity of interactions between the natural and digital worlds.