ARBD · AUTHOR REPER DATABASE · BUILD 28

From texts to an author’s concept system

ARBD is a new executable core for author corpus analysis in the KLT-RBDv6 project. It accumulates repers across an author’s works, preserves their provenance, and connects recurrence, grounds, tensions and gaps in a reviewable model.

Project author and author of the methodological formulation and claimed development direction: Ivan Borisovich Kurpishev. Author of the primary texts in the first corpus: Immanuel Kant.

Python 0.1.0KANT-B28-R017-AUDIT-001SCIENCE=ABSTAIN

How the author collection is organised

An occurrence is first located in a specific source, then linked to a reper candidate. Translations and new editions of the same work share one family and do not increase independent recurrence.

  1. 01

    Author

    Whose corpus is being studied

  2. 02

    Work family

    Editions and translations of one work

  3. 03

    Work / edition

    Language, version, source and domain

  4. 04

    Reper occurrence

    Locator, attribution, R/I/U/D/Dom

  5. 05

    Author reper candidate

    A reviewable group of occurrences

  6. 06

    Concept graph

    Relation types, provenance and time

The project’s claimed contribution connects the author reper collection with KLT-RBD: R/I/U/D, domain Dom, independent work families, attribution checks, a vector profile and a register of grounds. Completeness of the reconstructed concept system remains a research question.

What the engine does

Preserves attribution

Authorial text is distinguished from quotations of others, editorial voice and unverified attribution. Secondary literature does not populate the author’s own reper collection.

Reports the ACP profile

Seven coordinates: recurrence across families, time span, attribution, D and Dom coverage, contradiction load and argument gaps. No universal truth score is computed.

Locates tensions

A contradiction requires reviewed P and not-P with the same domain, modality, time and term sense. Without these conditions the result is ABSTAIN.

First corpus: Immanuel Kant

Inheriting audit KANT-B28-R017-AUDIT-001, 5 September 2026. ARBD 0.1.0 implements the Build 28 specifications and rechecks the preserved RUN-017.

works / editions
20
work families
14
reper occurrences
145
reper candidates
55
source-bound occurrences
109
core candidates
4

109 of 145 occurrences have reproducible source bindings; 106 of these have accepted authorial attribution, while three from the editorially mediated Logic require review. The package preserves 136 AA transcription pages. No complete work has been frozen.

87 of 90 quotation records passed locator checks under the declared normalisation rules. Three quotation records and three line ranges require repair. Provenance checks do not establish a proposition’s truth or the adequacy of its paraphrase.

Four structural core candidates

AUTHOR_CORE_CANDIDATE is an operational structural status on the current annotations. The module requires two independent families, admissible attribution, no direct unresolved contradiction within the checked fragment, non-overlapping authorial membership, and two families with reproducible sources. It does not establish truth or close every gap.

KANT-AR-035

Enlightenment and autonomous use of understanding

AUTHOR_CORE_CANDIDATE
Profile and sources
ACP coordinateValue
Independent families2
Years of works1784–1784
Attribution coverage2/2
Grounds D coverage2/2
Domain Dom coverage1/2
Direct unresolved contradictions in the checked fragment0
Gaps in registered argumentation0

Open diagnostic records: 4

AA source fragments

KANT-AR-039

Republican constitutionalism and freedom under law

AUTHOR_CORE_CANDIDATE
Profile and sources
ACP coordinateValue
Independent families2
Years of works1795–1798
Attribution coverage2/2
Grounds D coverage2/2
Domain Dom coverage1/2
Direct unresolved contradictions in the checked fragment0
Gaps in registered argumentation0

Open diagnostic records: 7

AA source fragments

Tensions and limits of inference

All five inherited tensions remain under review: nature and freedom; limits of cognition and practical reality; good and evil in the character of the species; revolution and reform; and criteria of evidence for progress. No case is presented as a proven direct contradiction or as finally resolved.

The 43 inherited semantic gaps and 287 new diagnostic records belong to different accounting levels. They cannot be added as a count of “Kant’s errors”. The narrow logical rule does not apply to any of the 1,711 pairs of inherited formalisations; consistency of the corpus is not established.

SCIENCE=ABSTAIN · truth_layer_promotion=0. Independent semantic review has not been performed.

Execution and connection to KLT-RBDv6

Python core 0.1.0, a JSON model, SQLite and a local HTML report are available. The published QA reports 34/34 software checks and an offline replay. This is a software-module trial on the supplied corpus.

An integration contract is prepared. This release did not deploy ARBD in the live analyser, import it into the production database, provide a browser/Flutter port or build an APK. The exhibition presents the checked module and its results; the existing analyser is linked separately.

  1. Source module ready
  2. Analyser adapter
  3. Staging import
  4. Integration QA
  5. Web / APK release
Open KLT-RBDv6

Primary materials and evidence navigation

Exhibition metrics come from ARBD_MODEL.json, SOURCE_AUDIT.json and QA_REPORT.json in the preserved package. Public links lead to the author’s canonical archive.

Machine-readable source ledger

Exhibition data