DATASETS FRONTIER · HOW IT WORKS

How it works

The detailed version of the home page: how a page is read, how a clause is logged, how the built copy is fixed, and how one measure is followed from edition to edition. The worked example below is real, and every quote is as printed.

01 / MAKE THE EVIDENCE VISIBLE

Quoted word for word, or marked as not verified

See the original page, the exact wording and the edition in force, side by side. If a match can't be verified, it stays marked unverified.

EVIDENCE VIEWFrom the page to the over-time library
Worked example
Ordinance 70 · Building (Reprint Certified 25 February 1975) · clause 27.4 · page 124one measure, hydrants, all four stages

  1. Read the page

    • Each page is read through a proprietary OCR process.
    • Disputed lines stay visible, never silently resolved.
    WHAT YOU SEE

    Clause 27.4 · Hydrants. Required in certain buildings.

    27.4. (1) One or more hydrants shall be provided at the level of every storey in the following buildings:

    (a) Class II, III, V, VI, VII and VIII buildings having a rise of more than four storeys, irrespective of flood area.

    (b) Class VI, VII, and VIII buildings in which, irrespective of the rise in storeys, the sum of the floor areas of all storeys exceeds the relevant floor area figure set out in the second column of Table 27.4.

    {{OCR? disputed: "Required in" | "buildings." }}

    Page 124: 51 of 66 lines confirmed; the 15 disputed lines stay marked like this one. The page prints "flood area"; it is quoted as printed.

  2. Log the clause

    • One record per source document, with its own document id.
    • Each clause is a tracked unit with a unit id and a hash of its text.
    • Every built page records where it came from.
    WHAT YOU SEE

    document id 09023151fd144ec6

    09023151fd144ec6/p124/27.4

    text_sha: example

    Document id and address as the Library records them; the hash shown is a stand-in.

  3. Fix the built copy

    • The source PDF is never written to.
    • Page furniture is removed from the built copy; every removal is logged and restorable.
    • After a fix, the affected pages are checked again.
    WHAT YOU SEE

    removed: running head · place: top of page · reason: page furniture

    Restorable

  4. Track over time

    • Editions are registered, with lineage across releases.
    • The same hash means unchanged; a different hash means reworded.
    • The reason for a change is never computed.
    WHAT YOU SEE
    1. 1975 reprint · clause 27.4text_sha: differs · reworded
    2. 1983 reprint · clause 27.4renumbered · 27.4 became 27.3
    3. 1988 reprint · clause 27.3

    The chain follows the measure, not the clause number.

    Reason as printed: Part 27 substituted by the amendment gazetted 1 June 1984 (Table of Amendments, 1992 reprint).

    Reprint dates as printed: 25 February 1975 · 1 June 1983 · 4 July 1988

EVOLUTION

The same measure, edition by edition

Worked example

Ordinance 70 (NSW, under the Local Government Act 1919). Each cut-out is the text as printed on its page. The Library follows the measure, not the clause number: when a later edition moves or renumbers a requirement, the chain still joins them.

Hydrants

1975 clause 27.4 → 1988 clause 27.3
1975 REPRINT

Ordinance 70 - Building (Reprint Certified 25 February 1975)

Clause 27.4 · Hydrants. Required in certain buildings.

Page 124

27.4. (1) One or more hydrants shall be provided at the level of every storey in the following buildings: (a) Class II, III, V, VI, VII and VIII buildings having a rise of more than four storeys, irrespective of flood area.

(2) In the buildings referred to in subclause (1), no part of a storey shall be outside a radius of 36 m from a hydrant installed within that storey.

09023151fd144ec6/p124/27.4

renumbered 27.4 → 27.3 · text_sha differs
1988 REPRINT

Ordinance 70 - Building (Reprint as at 4 July 1988)

Clause 27.3 · Hydrants. Installation required.

Pages 201 to 203

27.3. (1) Hydrants shall be installed in a building of a class specified in Column 1 of Table 27.3 in the locations specified opposite that class in Column 2 of that Table.

(3) Required hydrants— (a) shall be designed and installed in accordance with Ministerial Specification No. 10; (b) shall be connected to a fire main; … (d) shall be so situated that no part of the floor of any storey of the building … is more than— (i) where the building is not more than 25 m in height—36 m; or (ii) where the building is more than 25 m in height—30 m, from a hydrant installed in that storey.

5800ad1a3d42771a/p201/27.3

What changed:
  • Clause number: 27.4 became 27.3 when Part 27 was substituted (1 June 1984).
  • Building classes: the four-storey rule now also covers Classes IV and IX, set out in Table 27.3.
  • Floor area triggers: Class VII other cases and VIIIa rise from 2 800 to 3 000 m²; Class IX added.
  • Distance: one 36 m radius became 36 m up to 25 m high and 30 m above it.
  • New: designed to Ministerial Specification No. 10, connected to a fire main, within 3 m of a fire-isolated stair door, and a Board of Fire Commissioners report before installation.

Why the number moved: Contents, 1983: 27.3 Hose reels, 27.4 Hydrants. Contents, 1988: 27.2 Hose reels, 27.3 Hydrants. The 1984 substitution of Part 27 dropped clause 27.1 "Application of Part", so each fire-fighting clause moved up one number.

The 1983 reprint still numbers the hydrant clause 27.4 (page 185, 33a7b66998352c53/p185/27.4) and adds clause 12.3 "Fire main and hydrants. Initial installation." (page 48, 33a7b66998352c53/p48/12.3). The 1975 page prints "flood area"; it is quoted as printed.

Smoke alarms

NSW clause E1.7 (BCA 1990 Amendment 6) → BCA 1996 Volume Two clause 3.7.2.2
BCA 1990 · AMENDMENT 6 · NSW

Building Code of Australia 1990, Amendment 6

Clause NSW E1.7 · Fire and smoke alarms

Page 210

(c) In a Class 1 building or dwelling, (a) is satisfied by-

(i) a smoke alarm system complying with Specification E1.7; or

(ii) smoke alarms- (A) complying with AS 3786 or listed in the SSL Register of Accredited Products; and (B) where the building is provided with mains electrical power, connected to the mains and have a standby power supply; and (C) installed in suitable locations on or near the ceiling in (i) any storey containing bedrooms-

912c29d617119152/p210/NSW_E1.7

NSW E1.7 → 3.7.2.2 · text_sha differs
BCA 1996 · VOLUME TWO

Building Code of Australia 1996, Volume Two (Housing Provisions)

Clause 3.7.2.2 · Requirements for smoke alarms

Page 274

Smoke alarms must be installed in Class 1 buildings as follows:

(a) Installation requirements Smoke detection and alarm systems must be installed as follows: (i) In Class 1a buildings - consist of a system of smoke alarms located in accordance with 3.7.2.3; or (ii) in Class 1b buildings - consist of a system of smoke alarms- (A) located in accordance with 3.7.2.4; and (B) complying with Specification E2.2(a) of the BCA-Volume One.

(b) Design of smoke alarms Smoke alarms must comply with AS 3786.

(c) Smoke alarms must be connected to the consumer mains power where consumer power is supplied to the building.

a6098f6c3cd6ee82/p274/3.7.2.2

What changed:
  • Clause: the NSW variation E1.7 became a national clause, 3.7.2.2, in the housing volume.
  • Reach: a New South Wales rule became an Australia-wide rule; the 1996 clause excludes Class 1a houses in the Northern Territory.
  • Standard: AS 3786 or the SSL Register became AS 3786 only.
  • Power: "connected to the mains and have a standby power supply" became "connected to the consumer mains power"; the standby supply is no longer stated.
  • Locations: moved out of the clause into 3.7.2.3 for Class 1a houses and 3.7.2.4 for Class 1b buildings.

Amendment 6 states its purpose as the mandatory installation of automatic fire detection and alarm systems in Class 1a buildings or dwellings in NSW. Ordinance 70 has no smoke alarm clause of its own; its 1992 reprint adopts the Building Code of Australia (clause 1.3).

If the source can't be found, the result says so. Verification will be validated with pilot partners; results are drafts for a qualified person to review.

Read the verification standard

Ordinance 70 is New South Wales legislation (© State of New South Wales); the smoke alarm extracts are from the Building Code of Australia 1990 (Amendment 6, NSW) and 1996 (Volume Two). The extracts above are short passages reproduced from the owner's library copies to show the method; each was checked word for word against its page.

02 / ONE INFRASTRUCTURE, MANY LIBRARIES

Any text library that changes over time

Building codes and standards are the first library. The same template fits other text libraries.

01

Standards and building codes

The first library: standards and building codes, tracked edition by edition.

Bring
Standards, codes and edition history.
Review
Exact source text with its clause, edition and date.
Evaluate
Faithful extraction across every edition in the library.
Talk to us about this use case
02

Legal PDF libraries

A library of about 1,000 legal PDF files can be tracked the same way, document by document and version by version.

Bring
Legal PDF files and their version history.
Review
Exact source text with its document, section and version.
Evaluate
Faithful extraction across revisions and amendments.
Talk to us about this use case
03

Policies, specifications and contracts

Any text library that moves through editions fits the same template, inside the systems you already use.

Bring
Policies, specifications and contract documents.
Review
What changed, when it changed and where it applies.
Evaluate
Coverage and tracking across the library.
See the coach and services

The template is reproducible, so a new data set is designed to cost much less to stand up for the next library.

03 / COACH AND SERVICES

Understand the tracking, then design the workflow

The coach uses an LLM to explain how the tracking works and the key points an edition-over-time end point needs; it may read the files you choose to give it. Its answers are written by an LLM, so check any text it quotes against the source; the extraction that captures the source text uses no LLM. Around the infrastructure sit these services.

  1. Succinct reports

    Reports with LLM-assisted structuring, where the AI structures the report and the report is designed to keep the cited text verbatim from the source.

  2. JavaScript workflows

    Workflows and packages for audits and reporting, built to run beside the systems you already use.

  3. Bespoke and data services

    Bespoke workflows, plus data clean-up, recovery and integration for asset management software.