計算論的転換を読み解く：AI活用財務諸表分析における世界的潮流と異文化間の課題

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:AI×会計・ESG開示分析の研究潮流を5クラスターで整理した地図として、研究課題設定に活用できる。

🏢実務担当者:NLP/LLMによる開示文書処理の導入検討や、多言語・多制度下での開示対応コスト管理の示唆が得られる。

🏛政策担当者:機械可読開示の標準化と、言語・制度差がもたらす比較可能性の課題を規制設計に反映する視点を提供する。

Abstract Objective The paper aims to map the recent methodological reconfiguration of financial statement analysis (FSA) and to identify the analytical trajectories emerging between 2015 and 2025. It specifically examines how corporate disclosure is being reshaped by the computational turn, as it expands beyond structured financial data to include narrative, sustainability and machine-readable reporting across different cultural and regulatory contexts. Methodology A Scopus-based bibliometric design was adopted, applied to a corpus of 454 documents. It combines performance analysis, co-word analysis, thematic mapping and trend-topic analysis, processed through the Bibliometrix/ Biblioshiny environment to visualize the evolution of the field. Findings The results indicate that FSA is undergoing a profound rearticulation, moving from traditional ratio-based interpretation to an AI-enabled ecosystem. The evidence suggests a three-phase evolution: an initial decision-support phase, an intermediate expansion towards broader reporting architectures and non--financial disclosures and a recent computational turn centred on AI-enabled methods. The analysis reveals a five-cluster conceptual structure: (1) the traditional financial-statement core, (2) computational learning methods (ML/DL), (3) predictive applications in auditing and fraud detection, (4) sustainability and ESG extensions and (5) the cross-cultural dimensions of accounting research. Furthermore, the findings highlight a recent shift in which Artificial Intelligence (AI), natural language processing (NLP) and Large Language Models (LLMs) have become foundational to navigating global disclosure processing costs. Finally, the geography of collaboration reveals that methodological innovation in FSA is being shaped through specific cross-cultural research corridors, particularly the Sino–Anglophone and Asia–Pacific networks, while European participation currently lags behind regulatory developments such as CSRD/ESRS. Value Added The study contributes to the literature by conceptualizing the “computational turn” in FSA, offering an empirically grounded map of field’s methodological evolution. The paper argues that this transformation is a rearticulation rather than a substitution; new computational methods are being embedded within the traditional inferential logic of FSA rather than replacing it. Furthermore, the research integrates previously isolated streams – specifically NLP in accounting, machine learning in finance, ESG analytics and digital reporting – into a unified framework. A key contribution lies in the finding that while AI provides universal analytical power, cross-cultural challenges in the linguistic and contextual interpretation of narrative disclosures remain a critical frontier. By repositioning FSA as a domain where language, accounting tradition and institutional culture jointly dictate the credibility of computational measurements, this study opens a vital research interface between international management and global accounting practices. Recommendations Future research should develop integrative cross-stream studies that treat the five identified clusters as connected subsystems, with particular attention paid to the interface between computational methods and sustainability reporting. Domain-specific validation frameworks for ML applications in accounting and structured research programmes on LLM-based financial document analysis have been identified as priority agendas.

• openaire https://doi.org/10.2478/joim-2026-0001first seen 2026-11-05 04:47:52