KOTAX：韓国上場企業の財務・税務パネルに紐づく統合企業開示コーパス

KOTAX: An integrated corporate disclosure corpus linked to a financial and tax panel of Korean listed firms (原題)

Sejin Myung, Hyungjong Na, 서준일, Minsoo Lim, Hwa Su Hong, Hyungjoon Kim

Zenodo (CERN European Organization for Nuclear Research)データセット2026-09-17#開示インフラ対象セクター: finance

原典: https://doi.org/10.5281/zenodo.22806330

🤖 gxceed AI 要約

日本語

KOTAXはKOSPI・KOSDAQ上場企業の企業–年財務・税務パネルと、DART公示原本XMLを紐づけた大規模開示コーパスである。2011–2025年度の3種文書（事業報告・監査報告・連結監査報告）約6万件と、25,145行の企業×年パネル、60,902行の文書単位財務パネルを収録する。原本XMLを保持しつつ、出所追跡や検証結果、コードスナップショットをv1.1で追加し、開示テキスト分析の基盤を提供する。

English

KOTAX is a large-scale disclosure corpus linking a firm–year financial and tax panel of KOSPI/KOSDAQ-listed Korean firms to raw DART disclosure XML. It covers ~60,908 documents (business, audit, and consolidated audit reports, FY2011–2025), a 25,145-row firm–year panel, and a 60,902-row document-level financial panel. Version 1.1 adds provenance tracking, validation results, and code snapshots, offering a foundation for disclosure text analysis.

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・有報・統合報告書のテキスト分析基盤整備が課題であり、韓国のDARTを活用した本コーパスは、アジア圏の開示データ整備・比較研究の参考事例となる。日本版EDINETを用いた同種のコーパス構築や、日韓比較の開示研究への示唆がある。

In the global GX context

As global disclosure scholarship increasingly relies on machine-readable regulatory filings (EDGAR, ESEF/CSRD, EDINET), KOTAX demonstrates a reproducible pipeline for building linked financial–disclosure corpora from a national filing system (DART). It offers a methodological template for ISSB-aligned disclosure research and cross-jurisdiction comparison.

👥 読者別の含意

🔬研究者:開示テキストと財務データを紐づけた実証研究の基盤データセットとして活用できる。

🏢実務担当者:韓国企業の開示・財務情報を横断的に扱う際のデータソースとして参照可能。

🏛政策担当者:開示データの機械可読化・標準化と、それを支える公的インフラ設計の参考になる。

📄 Abstract（原文）

KOTAX는 KOSPI·KOSDAQ 상장기업의 기업–연도 재무·세무 패널과, 이에 연결된 DART 공시 원본 XML로 구성됩니다. 문서 내용은 전처리 전 원본 그대로이며, 파일명·폴더명만 전역 유일 영문 식별자로 표준화했습니다. DART_source_xml.zip (원본 XML): 2,256개 회사 폴더, 60,908개 파일, 회계연도 2011–2025, 문서 유형 3종(business_report 24,637 / audit_report 20,049 / consolidated_audit_report 16,222). dataset_with_xml.csv: 회사×연도 분석 패널, 25,145행, 51개 컬럼. KoTaP 구축 방법론을 따라 OpenDART 재무정보 API와 공공데이터포털(data.go.kr) 자료로 구축했으며(금액은 원 단위), 원본 XML 참조 컬럼 3개로 문서와 연결됩니다. financial_panel.csv: 공시 문서에서 공개 코드로 추출한 문서 단위 재무 패널, 60,902행, 27개 컬럼(총자산·총부채·매출·종업원 수·감사의견 및 단위 검증 플래그). company_name_changes.csv: 회사별 사명 변경 이력, 2,256행. document_manifest.csv (v1.1 신규): 문서별 출처 기록(DART 접수번호, 고유번호, 목록 조회일 2026-06-08, 원문 취득일). 기타 v1.1 신규 파일: financial_panel_v1.1_changelog.json, KOTAX_validation_v1.1.zip, KOTAX_code_v1.1.zip, LICENSES.md (버전 간 변경 행, 파서 독립 전수 검증 결과, 코드 스냅샷, 구성요소별 출처·이용조건). 주의사항: DART XML은 표준 XML이 아닌 전용 마크업으로, 커스텀 엔티티 &cr;를 파싱 전에 치환해야 합니다. financial_panel.csv의 매출 등은 기업 간 비교 전 단위를 반드시 확인해야 합니다. 상세 내용은 README.md를 참조하세요. [English] KOTAX consists of a firm–year financial and tax panel of KOSPI- and KOSDAQ-listed companies, linked to the original DART disclosure XML files. File contents are kept in their raw, unedited form; only file and directory names have been standardized to globally unique English identifiers. DART_source_xml.zip (Raw XML): 2,256 company folders, 60,908 files covering fiscal years 2011–2025 across three document types (24,637 business reports, 20,049 audit reports, and 16,222 consolidated audit reports). dataset_with_xml.csv: Firm–year analysis panel with 25,145 rows and 51 columns. Constructed following the KoTaP methodology using the OpenDART Financial Information API and Public Data Portal (data.go.kr) data (amounts in KRW). Linked to the raw XML via three document-reference columns. financial_panel.csv: Document-level financial panel extracted via open-source code from disclosures (60,902 rows, 27 columns: total assets, total liabilities, revenue, number of employees, audit opinions, and unit validation flags). company_name_changes.csv: Name-change history by firm (2,256 rows). document_manifest.csv (New in v1.1): Per-document provenance tracking (DART receipt number, corporation code, list retrieval date 2026-06-08, and source retrieval date). Other v1.1 Additions: financial_panel_v1.1_changelog.json, KOTAX_validation_v1.1.zip (parser-independent full-corpus validation results), KOTAX_code_v1.1.zip (code snapshot; also available at github.com/rydidcodru/KOTAX, tag v1.1), and LICENSES.md (component-level sources and terms of use). Caveats: DART XML uses a proprietary markup rather than standard XML; the custom entity &cr; must be substituted prior to parsing. Users should verify unit consistency (e.g., revenue in financial_panel.csv) before making cross-firm comparisons. See README.md for details.

🔗 Provenance — このレコードを発見したソース

openalex https://doi.org/10.5281/zenodo.21526231first seen 2026-09-19 05:11:57 · last seen 2026-09-19 05:12:05

🔔 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。