Datasets:
configs:
- config_name: principal_legislation
data_files:
- split: train
path: frl_principal_legislation_all.jsonl
- config_name: amendment_history
data_files:
- split: train
path: frl_legislation_history_all.jsonl
- config_name: legislation_with_history
data_files:
- split: train
path: frl_legislation_with_history_all.jsonl
tags:
- law
- tax
- australia
- legislation
- statutes
- legal
- rag
- nlp
- legal-nlp
- amendment-history
- legislative-history
- itaa1997
- itaa1936
- gst
- domain-specific
- text-retrieval
- question-answering
- fine-tuning
language:
- en
license: cc-by-4.0
pretty_name: Australian Tax Legislation & Amendment History — ITAA 1997, ITAA 1936, GST Act
size_categories:
- 10K<n<100K
Australian Tax Legislation & Amendment History
The full text of every section of the 22 principal Acts the Australian Taxation Office administers — Income Tax Assessment Act 1997, Income Tax Assessment Act 1936, the GST Act, FBTAA, the Taxation Administration Act 1953, the superannuation and fuel tax Acts and more — each section joined to its complete amendment history: which Act changed it, which schedule item, and when it commenced. Sourced from the Federal Register of Legislation and structured for legal RAG, statutory retrieval, and LLM fine-tuning.
This rebuilds the "see history notes" view that commercial Australian tax services charge for — reconstructed entirely from the freely licensed, Crown-copyright FRL compilation endnotes, so it can be published under CC BY 4.0.
Companion dataset: ATO rulings, private advice, determinations and guidance are published separately at simplelex/ATO-Australian-Tax-Rulings-and-Guidance. The two are designed to be used together — ATO documents cite provisions; this dataset resolves them.
File formats
Each dataset is published in two parallel files with identical content and field order:
| Format | Filename pattern | Best for |
|---|---|---|
| CSV (UTF-8) | <dataset>.csv |
Pandas, spreadsheets, SQL bulk-load |
| NDJSON / JSON Lines | <dataset>.jsonl |
datasets.load_dataset(), streaming, jq, DuckDB, RAG ingestion |
JSON keys are identical to CSV column names — the schema tables below describe both formats.
Contents
- Dataset Splits
- Quick Start
- Schema
- Acts Covered
- Use Cases
- Limitations
- Attribution — required wording
- Source & Licensing
- Citation
- Terms of Service
Dataset Splits
| Split | Content | Est. Rows | Coverage | CSV file | NDJSON file |
|---|---|---|---|---|---|
principal_legislation |
Current in-force text of every section of the 22 ATO-administered Acts | ~10,000 sections | Latest compilation | frl_principal_legislation_all.csv |
frl_principal_legislation_all.jsonl |
amendment_history |
One row per amendment event — provision, action, amending Act, commencement | ~43,000 events | 1986–present | frl_legislation_history_all.csv |
frl_legislation_history_all.jsonl |
legislation_with_history |
Flagship. Each current section joined to its chronological amendment history | ~10,000 sections | Current + history | frl_legislation_with_history_all.csv |
frl_legislation_with_history_all.jsonl |
Quick Start
Option A — NDJSON via datasets (recommended for ML)
from datasets import load_dataset
ds = load_dataset(
"simplelex/Australian-Tax-Legislation-and-Amendment-History",
"legislation_with_history",
split="train",
)
print(ds[0])
# Available configs: principal_legislation · amendment_history · legislation_with_history
Option B — NDJSON via pandas
import pandas as pd
df = pd.read_json("frl_legislation_with_history_all.jsonl", lines=True)
# What does ITAA 1997 s 6-5 say, and how has it changed?
s6_5 = df[(df["Act_Short_Name"] == "ITAA1997") & (df["Provision_Key"] == "s6-5")]
print(s6_5[["Heading", "Text", "Amendment_Count", "History_Notes"]].iloc[0])
Option C — join provisions to their amendment events yourself
import pandas as pd
provisions = pd.read_json("frl_principal_legislation_all.jsonl", lines=True)
events = pd.read_json("frl_legislation_history_all.jsonl", lines=True)
joined = provisions.merge(events, on=["Act_FRL_Id", "Provision_Key"], how="left")
# Most-amended sections of the ITAA 1997
top = (
joined[joined["Act_Short_Name_x"] == "ITAA1997"]
.groupby(["Provision_x", "Heading"])
.size()
.sort_values(ascending=False)
)
print(top.head(10))
Multi-value fields (Amending_Acts, History_Notes) use | as the internal delimiter in both formats.
Schema
The column tables below apply to both the CSV and NDJSON variants — every CSV column maps 1:1 to a JSON key of the same name in the corresponding .jsonl file, in the same order.
Encoding conventions
| Convention | Detail |
|---|---|
| Multi-value delimiter | | (space-pipe-space) within a single cell / string field |
| Empty values | Empty string "" — never NULL, N/A, or JSON null |
| Dates | Verbatim as published by the FRL — no normalisation |
| Defined terms | * prefixes a defined term in provision text (the FRL's own convention, preserved) |
1. Principal Legislation (principal_legislation)
Current in-force text of every section of the 22 principal Acts the ATO administers. One row per provision.
| Column | Type | Nullable | Description |
|---|---|---|---|
Act_Short_Name |
string | No | Short code, e.g. ITAA1997, FBTAA, GST. |
Act_Title |
string | No | Full Act title, e.g. Income Tax Assessment Act 1997. |
Act_Year |
int | No | Year of the Act. |
Act_FRL_Id |
string | No | FRL series id, e.g. C2004A05138. Note this is not the compilation register id. |
Compilation_PiT |
string | No | Compilation point-in-time; latest for the current compilation. |
Provision |
string | No | Section label, e.g. s 6-5, s 136, s 102UQ. |
Provision_Key |
string | No | Normalised join key to the history datasets (s 6-5→s6-5, Section 3→s3; sub-provisions roll up to their section). |
Heading |
string | Yes | Section heading, e.g. Income according to ordinary concepts (ordinary income). |
Text |
string | Yes | Full operative text (subsections/notes/tables concatenated). Genuinely empty for reserved and "Map of…" navigational sections. |
Source_URL |
string | No | https://www.legislation.gov.au/{Act_FRL_Id}/latest/text#{provision}. Resume key. |
2. Amendment History (amendment_history)
One row per amendment event: which provision changed, the action taken, the amending Act, and when it commenced. Reconstructed from FRL compilation endnotes (Endnote 2 abbreviation key, Endnote 3 legislation history, Endnote 4 amendment history).
| Column | Type | Nullable | Description |
|---|---|---|---|
Act_Short_Name / Act_Title / Act_Year / Act_FRL_Id |
— | No | Identify the principal Act. |
Provision |
string | No | Provision affected, e.g. s 5B, s. 41-5(1), Note 1 to s. 41-5(1). |
Provision_Key |
string | No | Normalised key; joins to principal_legislation. Sub-provision and note amendments roll up to their section. |
Action_Code |
string | Yes | Raw endnote code: am amended, ad added/inserted, rep repealed, rs repealed & substituted. |
Action |
string | No | Expanded verb, e.g. Amended, Inserted, Repealed. |
Amending_Act_Number |
int | No | Number of the amending Act. |
Amending_Act_Year |
int | No | Year of the amending Act. |
Amending_Act_FRL_Id |
string | Yes | FRL id of the amending Act, e.g. C2013A00124. |
Amending_Item |
string | Yes | Schedule item that made the change, e.g. Sch 11 item 10 (best-effort; ~58% filled — see Limitations). |
Commencement |
string | Yes | Commencement detail, e.g. Sch 11 (items 10-25, 27): 30 June 2013. |
Application_Saving_Transitional |
string | Yes | Application/saving/transitional provisions where present. |
History_Note |
string | No | Reconstructed note, e.g. Amended by No 124 of 2013, effective 30 June 2013. Our wording, not the FRL's — see Attribution. |
Source_URL |
string | No | Stable per-event URL (resume key). |
3. Legislation + History Notes (legislation_with_history)
The flagship dataset. A left-join of #1 and #2 on (Act_FRL_Id, Provision_Key) — one row per current section, carrying its full text AND its complete chronological amendment history.
| Column | Type | Nullable | Description |
|---|---|---|---|
Act_Short_Name / Act_Title / Act_Year / Act_FRL_Id |
— | No | Identify the principal Act. |
Provision / Provision_Key / Heading / Text |
— | — | The current section text (as in #1). |
Amendment_Count |
int | No | Number of amendment events (0 = never amended since the Act commenced). |
First_Amended |
string | Yes | Earliest amending Act, e.g. No 139 of 1987. |
Last_Amended |
string | Yes | Most recent amending Act. |
Amending_Acts |
string | Yes | Distinct amending Acts in chronological order, |-separated. |
History_Notes |
string | Yes | Chronological reconstructed notes: Amended by No 139 of 1987, effective 18 Dec 1987 | Amended by No 145 of 1995, …. |
Source_URL |
string | No | The provision's /latest/text#{provision} URL. |
Acts Covered
The 22 principal Acts administered by the Australian Taxation Office, including:
- Income Tax Assessment Act 1997 and Income Tax Assessment Act 1936
- A New Tax System (Goods and Services Tax) Act 1999, plus the Luxury Car Tax and Wine Equalisation Tax Acts
- Fringe Benefits Tax Assessment Act 1986
- Taxation Administration Act 1953 (including the Schedule 1 administrative provisions)
- Superannuation Industry (Supervision) Act 1993, Superannuation Guarantee (Administration) Act 1992 and related superannuation Acts
- Fuel Tax Act 2006, Petroleum Resource Rent Tax Assessment Act 1987
- Tax Agent Services Act 2009
Use Cases
- Statutory RAG — each provision is a self-contained retrieval unit with a stable
Source_URL; retrieve the section text and its amendment lineage together so generated answers cite both the current provision and when it last changed - Temporal legislation models — train or evaluate on dated amendment events (
Commencement,Amending_Act_Year,Action) for legislative change prediction, version diffing, or "law as at date X" retrieval - Legislative graph construction — build provision-level change networks: which amending Acts touch which parts of the ITAA 1997, and how amendment activity clusters over time
- Fine-tuning Australian tax LLMs — the full operative text of the core tax statutes, section-segmented and cleanly labelled, as a grounding corpus
- Cross-referencing ATO guidance — explode
Relevant_Legislative_Provisionsin the companion ATO dataset and join onProvision_Keyto see every ruling that interprets a given section - Compliance and drafting tooling — surface the current text of a provision plus what changed and when, without a commercial subscription
Limitations
- Currency — compilations are checked for amendments automatically and the datasets rebuilt when a new compilation is registered. Between checks, a recently commenced amendment may not yet be reflected. Always verify against legislation.gov.au for official or time-sensitive purposes.
Amending_Itemis ~58% filled — the remaining blanks are genuinely unattributable from the endnotes: inserted sections that no schedule item names individually, and Acts published in the older OLDF endnote format (notably the Superannuation Guarantee Charge Act).- History notes are reconstructed, not reproduced —
History_NoteandHistory_Notesare generated in our own wording from the endnote facts (provision, action, amending Act, commencement). They are not the ATO Legal Database's history notes, which are separately copyrighted. - Repealed provisions — the joined dataset covers sections present in the current compilation. Provisions repealed or renumbered out of existence appear in
amendment_historybut have no row inprincipal_legislation, so they drop out of the left join. - Text extraction — provision text is extracted from the FRL's compilation epub files. Complex tables, formulae, and diagrams may be flattened or lost; check
Source_URLfor anything structurally unusual. - Point-in-time text is not in this dataset — per-compilation-version provision text (what any section said on any past date) exists but is served separately and is not published here.
- Australian Commonwealth jurisdiction only — no state or territory legislation, and no non-tax Commonwealth Acts.
- Not legal advice — see Terms of Service.
Attribution — required wording
The Federal Register of Legislation licence requires attribution in specific words, not just a credit. If you redistribute this data or content derived from it, use the statement that matches what you are redistributing:
For provision text reproduced substantially unchanged (the Text and Heading columns):
Sourced from the Federal Register of Legislation at [full date of download]. For the latest information on Australian Government law please go to https://www.legislation.gov.au.
For the reconstructed history notes and any restructured or derived content (History_Note, History_Notes, Amendment_Count, and the joined dataset generally):
Based on content from the Federal Register of Legislation at [full date of download]. For the latest information on Australian Government legislation please go to https://www.legislation.gov.au.
Substitute the date you downloaded the data — for this dataset, the "Last updated" date shown on this repo. Most users of the joined dataset need the "Based on" form, since the amendment history is derived rather than reproduced.
Source & Licensing
Source: Federal Register of Legislation (legislation.gov.au), © Commonwealth of Australia.
All content on the Federal Register of Legislation is provided under the Creative Commons Attribution 4.0 International licence (CC BY 4.0), with the exception of the Commonwealth Coat of Arms and any material identified as third-party copyright. Neither is reproduced in this dataset.
The licence permits commercial use, redistribution and adaptation, provided the attribution requirements above are met. This structured compilation is published by SimpleLex under the same CC BY 4.0 terms.
The amendment-history notes are reconstructed in our own wording from the FRL's Crown-copyright compilation endnotes — deliberately, so that no third-party commercial history notes are reproduced.
Citation
@dataset{simplelex_au_tax_legislation_2026,
author = {SimpleLex},
title = {Australian Tax Legislation \& Amendment History},
year = {2026},
publisher = {Hugging Face},
url = {https://huggingface.co/datasets/simplelex/Australian-Tax-Legislation-and-Amendment-History},
note = {Based on content from the Federal Register of Legislation,
https://www.legislation.gov.au, CC BY 4.0}
}
Terms of Service
Last updated: July 2026
What this product is
A structured dataset of the principal Acts administered by the Australian Taxation Office, together with reconstructed per-section amendment histories, sourced from the Federal Register of Legislation (legislation.gov.au) and compiled into machine-readable CSV and NDJSON format.
Permitted use
You may use this dataset for any lawful purpose, including research, analysis, building applications, and commercial projects. You do not need to ask permission. You do need to carry the FRL attribution wording set out above when redistributing the source content — that requirement comes from the Commonwealth's licence, not from us.
Source material and copyright
The underlying legislation is © Commonwealth of Australia, published under the Creative Commons Attribution 4.0 licence, excluding the Commonwealth Coat of Arms and any third-party copyright material. The Commonwealth retains copyright in the original materials. This dataset is a structured compilation and does not claim copyright in the original source material.
Not legal or tax advice
Nothing in this dataset constitutes legal or tax advice. It is a reproduction and restructuring of publicly available legislation, provided for informational purposes only. Always consult a registered tax agent or legal professional for advice about your specific situation.
Accuracy and currency
We make no warranties about the completeness, accuracy, or currency of the data. Legislation is amended continuously and a compilation may be superseded at any time. The authoritative text is the one published at legislation.gov.au — always verify there for official or time-sensitive purposes.
No liability
To the maximum extent permitted by Australian law, simplelex accepts no liability for any direct, indirect, or consequential loss arising from use of or reliance on this dataset.
Changes
We may update these terms at any time. Continued use of the dataset following an update constitutes acceptance of the revised terms.
Support the Project
This dataset is free. If it saves you time on data engineering, a coffee is appreciated: