Method

How Abe Activity Graph is built, what it covers, and the limits readers should keep in mind.

1. Project scope

本プロジェクトは、第2期安倍政権期（2012年12月26日〜2020年9月16日）の 公開された首相動静を構造化し、組織別・カテゴリ別・ 月次推移の観点から分析するための研究用データ基盤です。

The project covers publicly recorded prime ministerial activity during the second Abe administration (2012-12-26 → 2020-09-16). The unit of analysis is the event row as it appears in the published daily-schedule record.

2. Principal model

安倍晋三 is the principal of every record in scope: the person whose calendar the daily-schedule documents. The principal is therefore tracked separately as a coverage axis, and is not counted as one of his own recorded contacts.

3. Public-record observation limit

本データは、首相動静に記録された公開上の接触を対象とする。そのため、 官邸内部で日常的に行われる随時協議、電話連絡、非公開調整、秘書官・ 補佐官を介した事前調整など、公開記録に現れない情報は含まれない。

This dataset covers only contacts recorded in the published daily-schedule. Routine in-office consultation, phone calls, non-public coordination, and preparatory adjustments mediated by secretaries and aides — none of which surface in the press-recorded log — are not included.

4. Data coverage

Both segments are normalized into a common derived-data schema, but source differences remain and affect comparability.

5. Comparability

現時点では、組織・カテゴリ・月次推移の比較が最も信頼できる分析単位です。

At this stage, organization-level, category-level, and monthly trend comparisons are the most reliable. Cross-segment person-level comparisons require additional care.

6. Person-level caveat

Recorded contacts are resolved canonical contacts only — i.e., names that the resolver matched against the curated canonical-person dictionary, including date-aware role-only matches.

• This is not a complete ranking of all people Abe met. People who do not yet have a canonical record, or whose mention appeared only as a surname-only fragment, remain in the unresolved diagnostic.
• Early-period person resolution remains incomplete. A genuine pre-2014-12 contact may be absent from the recorded-contact column simply because the underlying mention was a partial fragment.
• The principal (安倍晋三) is excluded from every contact ranking by construction.

7. Raw text policy

• Raw source text is not displayed. Public artifacts contain structured factual fields, source metadata, and SHA-256 hashes of the parsed segments for chain-of-custody.
• Raw newspaper and structured activity-log text remain local-only under data/raw/ (gitignored). A safety walk fails the build if any forbidden key (event_text, body, raw_text, description,verbatim_text, etc.) appears at any depth of a public artifact.
• Source URLs are preserved so a reader can navigate to the publisher and consult the original on the publisher’s terms.

8. Technical appendix

Internal identifiers used by the pipeline. These names do not appear on the public pages; they are listed here to help researchers reproducing or extending the analysis.

• early_abe_public_archive — internal layer name for the 2012-2014 public archive segment.
• kanteilog_daily_like — internal layer name for the 2014-2020 structured activity-log segment.
• COMPARABLE_ORG_LEVEL_ONLY — internal comparability verdict from scripts/analyze/abe_full_period_comparability_audit.py. The public surface presents the equivalent reader-facing wording in §5 above.
• prs-abe-shinzo — canonical principal_person_id used throughout the resolver and audit chain.

Citation

土屋貴裕『Abe Activity Graph』, version 0.1, accessed 2026-05-15.

Tsuchiya, Takahiro. Abe Activity Graph, version 0.1, accessed 2026-05-15.

Abe Activity Graph is a personal research project by Takahiro Tsuchiya (土屋貴裕). Not affiliated with the Kantei, the Cabinet Office, or any government body.