Sustainalyze: AI Powered Transperancy in ESG Reporting
Sustainalyze: ESG報告におけるAIを活用した透明性 (AI 翻訳)
Kunj Modh, Varun Patel, Vivek Devani
🤖 gxceed AI 要約
日本語
本論文は、ESG報告書の自動分析のためのAIプラットフォーム「ESG Analytics」を提案する。RAGアプローチを用いてテキスト・表形式データからESG指標を抽出し、FinBERT-ESGで分類、LLMで文脈に応じた抽出を行う。リスクスコアとインパクトスコアの二重評価、グリーンウォッシュ検出機能を備え、透明性とスケーラビリティを向上させる。
English
This paper presents ESG Analytics, an AI platform for automated analysis of ESG reports. It uses a Retrieval-Augmented Generation (RAG) approach to extract ESG metrics from text and tables, integrating FinBERT-ESG for classification, ChromaDB for semantic retrieval, and LLMs for context-aware extraction. The platform generates dual ESG scores (Risk and Impact) and includes a multi-signal greenwashing detection mechanism, improving transparency and scalability.
Unofficial AI-generated summary based on the public title and abstract. Not an official translation.
📝 gxceed 編集解説 — Why this matters
日本のGX文脈において
日本ではSSBJ基準や有価証券報告書でのESG開示が進む中、本プラットフォームは企業の開示データの自動分析・グリーンウォッシュ検出を可能にし、投資家対応や統合報告書の品質向上に貢献する。
In the global GX context
With global frameworks like TCFD, ISSB, and CSRD driving ESG disclosure, this AI platform addresses the need for automated, transparent analysis of large ESG reports. Its greenwashing detection adds a layer of accountability, supporting evidence-based decision-making for investors and regulators.
👥 読者別の含意
🔬研究者:The paper demonstrates a novel application of RAG and FinBERT for ESG metric extraction, offering a foundation for further research in AI-driven sustainability analytics.
🏢実務担当者:Corporate sustainability teams can use this platform to efficiently analyze ESG reports, benchmark performance, and detect potential greenwashing in their own disclosures.
🏛政策担当者:Regulators can leverage the platform's capabilities for oversight of ESG disclosures and to develop standards for automated reporting verification.
📄 Abstract（原文）
This work presents ESG Analytics, an AI-powered platform for automated analysis of Environmental, Social, and Governance (ESG) reports. ESG reports are often large, complex, and unstructured, making manual analysis inefficient and time-consuming. The proposed system uses a Retrieval-Augmented Generation (RAG) approach to extract ESG metrics from both textual and tabular data. It integrates FinBERT-ESG for classification, ChromaDB for semantic retrieval, and Large Language Models (LLMs) for accurate and context-aware extraction. The platform generates dual ESG scores, including Risk Score and Impact Score, and incorporates a multi-signal greenwashing detection mechanism to identify misleading sustainability claims. This system improves accuracy, transparency, and scalability in ESG analysis and provides evidence-based insights for better decision-making.
🔗 Provenance — このレコードを発見したソース
- openalex https://doi.org/10.5281/zenodo.19934800first seen 2026-05-05 21:28:55
gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。