企業サステナビリティ分析のための機械学習・テキストマイニング応用フレームワーク：NIFTY 50年次報告書（2025–26）研究

AN APPLIED MACHINE LEARNING AND TEXT MINING FRAMEWORK FOR CORPORATE SUSTAINABILITY ANALYSIS: A STUDY OF NIFTY FIFTY ANNUAL REPORTS (2025–26) (原題)

Sonia Kundra, Mandeep Kaur, Parwinder Kaur Dhillon, Dr. Neha Monga, Ashish Arora, Sandeep Chaudhary, Daljit Singh, Ikasdeep Singh Sahota

Acta Innovations📚 査読済 / ジャーナル2026-10-30#AI×ESG経営インパクト: 調達リスク対象セクター: cross_sector

DOI: 10.62441/actainnovations.vi.2832

原典: https://doi.org/10.62441/actainnovations.vi.2832

🤖 gxceed AI 要約

日本語

インドNifty 50の50社の年次報告書（BRSR・統合報告書含む）を対象に、機械学習とテキストマイニングを用いて6次元のサステナビリティ開示を分析した研究。報告とガバナンスの開示量が最多で、バリューチェーンと第三者保証が最少だった。環境集約型セクターは環境開示が有意に高く、社会・ガバナンス開示はセクター不変であることを示した。再現可能な開示指標SDIDIを提案する。

English

This study applies machine learning and text mining to the annual reports (including BRSR and integrated reports) of the 50 Nifty 50 companies in India, scoring disclosures across six dimensions. Reporting and Governance show the highest disclosure volumes, while Value Chain and Assurance are lowest. Environmentally intensive sectors disclose significantly more environmental information, whereas Social and Governance disclosures are sector-invariant. The authors propose a replicable Sustainable Development Initiative Disclosure Index (SDIDI).

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

インドBRSRを対象とした開示テキストの機械的分析手法は、SSBJ基準・有価証券報告書のサステナビリティ開示の自動評価や保証範囲の検証に応用可能で、日本企業・投資家にとって示唆が大きい。

In the global GX context

As mandatory sustainability disclosure expands under ISSB/CSRD and India's BRSR, this framework offers a replicable, ML-based method for benchmarking disclosure quality and assurance gaps across markets, informing global disclosure scholarship on sectoral divergence.

👥 読者別の含意

🔬研究者:ML・テキストマイニングによる開示スコアリングの再現可能な手法と、セクター間分散の統計的検証を参考にできる。

🏢実務担当者:バリューチェーン開示と第三者保証が相対的に手薄である点は、自社の開示ギャップ特定と保証対応の優先順位づけに活用できる。

🏛政策担当者:開示義務化後の実効性評価や、保証・バリューチェーン開示の規制強化の根拠として参照できる。

📄 Abstract（原文）

India’s transition to a growth model that is innovative, inclusive, and environmentally sustainable depends substantially on the conduct of its largest listed enterprises. This study examines the sustainable development initiatives and disclosures of the 50 constituents of the NSE Nifty 50 index, using their Annual Reports (including the Business Responsibility and Sustainability Report, BRSR, and Integrated Reports) for the financial year 2025–26 as the primary data source. Following the content analysis and relative disclosure score methodology of Azevedo et al. (2025), a directed content analysis was carried out across six core dimensionsReporting Framework (Reporting), Environmental & Climate Action (Environment), Social & Community Development (Social), Corporate Governance (Governance), Value-Chain Sustainability (Value_Chain), and Independent Assurance (Assurance)supplemented by a rescaled 0–100 Sustainable Development Initiative Disclosure Index (SDIDI). Across all 50 companies (N = 50), total sustainability disclosure counts averaged 844.02 (median = 792.00, SD = 404.51, range = 160 to 1,984), corresponding to a mean Relative Disclosure Index (RDI) of 42.54% of the sample benchmark and a mean equal-weighted Composite SDIDI of 37.87%. In raw volume terms, Reporting (mean = 220.08, 26.08% share) and Governance (mean = 218.66, 25.91% share) recorded the highest average disclosures, whereas Value_Chain (mean = 57.44, 6.81% share) and Assurance (mean = 42.08, 4.99% share) recorded the lowest; on a normalized 0–100 benchmark scale, Governance ranked highest (52.06%) and Assurance ranked second (48.93%), while Reporting (24.18%) and Environment (26.02%) showed the widest inter-firm dispersion. Kruskal–Wallis tests across eight sectors confirmed statistically significant sectoral differences in Environment (H = 15.68, p = 0.0282), Value_Chain (H = 18.30, p = 0.0107), and TOTAL disclosure (H = 14.57, p = 0.0419), with Mann–Whitney U tests confirming that environmentally intensive sectors record significantly higher environmental disclosures (U = 436.5, p = 0.0071) than other sectors, whereas Social (H = 9.19, p = 0.2390) and Governance (H = 9.95, p = 0.1913) disclosures are sector-invariant. The paper contributes a replicable disclosure-based measure of corporate sustainability action for the Indian large-cap universe and offers implications for regulators, investors, and policymakers

🔗 Provenance — このレコードを発見したソース

openalex https://doi.org/10.62441/actainnovations.vi.2832first seen 2026-11-04 05:06:16 · last seen 2026-11-06 05:02:19

🔔 こうした論文の新着を逃したくない方はキーワードアラートに登録（無料・3キーワードまで）。

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。