Knowledge Versioning — References

Temporal data models, snapshot isolation over large content stores, and the records a dispute or an audit actually needs.

Temporal and bitemporal models

• Developing Time-Oriented Database Applications in SQL — Richard T. Snodgrass, Morgan Kaufmann,
  1999. Valid time and transaction time, and the distinction between when something was true and when the system recorded it, which is the distinction a corpus revision needs. Cited by name.
• Temporal Databases: Theory, Design, and Implementation — Abdullah Uz Tansel, James Clifford, Shashi Gadia, Sushil Jajodia, Arie Segev and Richard Snodgrass (editors), Benjamin/Cummings,
  1993. The foundational treatment of time-varying data and the query semantics over it. Cited by name.
• ISO/IEC 9075:2011 (SQL:2011), temporal features — the standardised system-versioned and application-time period tables, for the structured side of a versioned corpus. Cited by name.

Snapshots over large content stores

• Apache Iceberg: table format specification — the snapshot and manifest model that gives a large table an identifiable state and a time-travel query. Cited by name; the specification is published with the project.
• A Critique of ANSI SQL Isolation Levels — Hal Berenson, Phil Bernstein, Jim Gray, Jim Melton, Elizabeth O'Neil and Patrick O'Neil, ACM SIGMOD, 1995. Snapshot isolation, and the precise sense in which a reader sees one consistent state of a changing store. Cited by name.

Why the record has to be joinable to the answer

• PROV-O: The PROV Ontology — the vocabulary for recording which activity, using which inputs, produced a given output, which is the shape of the join between an answer and a corpus revision.
• NIST SP 800-92: Guide to Computer Security Log Management — retention, integrity and correlation requirements for records that are expected to answer a question months later.

No verified reference is recorded for composite corpus revision identifiers, for splitting retention between explanation and reproduction, or for reconciling erasure obligations with retained index generations. Temporal database theory addresses facts rather than vectors, and the storage economics have no published treatment.