Bulk Data

Everything on this site is built from two flat datasets: 821 Illinois FOIA/OMA Public Access Opinions and PAC determination letters, and 287 Illinois appellate and Supreme Court cases that address FOIA or the Open Meetings Act. Both are free to download and reuse below — no account, no API key, no rate limit.

See Sources & Methods for what these documents are, where they came from, and how their summaries were generated and checked — read that before treating any field below as more authoritative than it is.

Downloads

Download foia-corpus.json Download caselaw.json

foia-corpus.json is the export of the 821-document PAC opinions/letters corpus. caselaw.json is the export of the 287-case appellate/Supreme Court dataset. Every case in it is one this site publishes.

Use the corpus with your own LLM

No hosted model or API key is required. Download the metadata export and the full-text JSONL shards, then load them into the local model, notebook, or research tool you already use. Each line is one document and includes the source text, structured fields, a stable site URL, and the original source URL.

Download full-text manifest Download metadata export

Full-text shards:01.jsonl · 02.jsonl · 03.jsonl · 04.jsonl

Caselaw full-text:manifest · 01.jsonl · 02.jsonl · 03.jsonl · 04.jsonl

Suggested workflow: retrieve by full_text and metadata, answer only from the returned source passages, link each claim to site_url or source_url, and treat needs_review=1 rows as requiring direct PDF verification. Generated summaries are convenience fields, not the legal source.

Use the Illinois FOIA/OMA corpus as a retrieval source. For every answer:
1. Prefer full_text and the original PDF over generated summaries.
2. Separate what the document says from your own inference.
3. Cite the document's site_url and source_url.
4. Call out needs_review=1, incomplete text, or conflicting fields.
5. Do not treat one PAC determination as a universal statement of law.

Full document text is not included in foia-corpus.json. The underlying build data carries a full_text field per document that alone runs to roughly 14MB across the corpus; it's stripped from this download to keep the file lean. Full text for any individual document is available on that document's own page on this site (/opinions/<id>) and, for court cases, at /case-text/<id>.txt— and the original source PDF, linked from every document page, is always the authoritative copy.

Schema — `foia-corpus.json`

One object per Public Access Opinion, PAC determination letter, or historical AG opinion.

Field	Meaning
`id`	Numeric document ID; also the URL slug at `/opinions/<id>`.
`doc_type`	"Public Access Opinion", "PAC Determination Letter", or an older AG opinion type.
`binding`	1 if this is a binding Public Access Opinion, 0 otherwise (determination letters and older opinions are non-binding).
`year` / `date_issued`	Year and, where recorded, exact issue date. Not every document has `date_issued`.
`case_number`	The PAC opinion number, where applicable.
`request_numbers`	Array of PAC Request for Review number(s) the document resolves.
`title` / `filename` / `source_url`	Display title, original source filename, and (where known) the URL it was scraped from.
`public_body` / `requesting_party`	The government body involved and the requester, as named in the document.
`primary_statute` / `foia_or_oma`	Which statute(s) the document addresses — FOIA (5 ILCS 140), OMA (5 ILCS 120), or both.
`foia_exemption`	FOIA exemption citation(s) discussed, where applicable. Stored inconsistently — some short-form ("7(1)(c)"), some full-citation — see the note below.
`disposition`	Clean enum outcome: `violation_found`, `no_violation`, `partial_violation`, `closed_no_further_action`, or `insufficient_info`. Use this field, not `pac_outcome`, for any filtering.
`pac_outcome`	Free-text outcome description. More specific than `disposition` in places, but has dozens of near-duplicate phrasings of the same handful of outcomes — display text only, not a clean filter field.
`summary` / `plain_summary` / `holding` / `legal_question`	Machine-generated summary fields for most documents; hand-written for the binding opinions — see Sources & Methods for how these were produced and checked. Not written by the original document's author.
`subject_category` / `faq_tags`	Topic classification and cross-reference tags used to group related documents on the site.
`ocr_status` / `ocr_quality`	Whether the document was OCR'd from a scan and, if so, an estimated confidence score.
`needs_review`	1 if this document's generated summary is still flagged as unverified against its source text (see Sources & Methods, section 3) — treat the source PDF as authoritative for these.
`splice_suspect`	1 if the extracted text was flagged as possibly containing a fragment of an unrelated document.
`word_count`	Word count of the extracted source text.

Schema — `caselaw.json`

One object per appellate or Supreme Court case that mentions FOIA or OMA.

Field	Meaning
`dedup_key`	Stable unique identifier for the case; also the URL slug at `/cases/<dedup_key>` (URL-encoded).
`neutral_citation` / `case_name`	The case's neutral citation (e.g. "2022 IL App (1st) 191101") and name.
`court_level` / `appellate_district`	Appellate or Supreme Court, and which appellate district, where applicable.
`year` / `county`	Decision year and originating county.
`appellate_docket` / `circuit_docket`	Docket numbers at each court level, where known.
`published`	1 if formally published, 0 if an unpublished/Rule 23 order.
`authoring_justice`	Justice credited with authoring the opinion, where known.
`statute` / `exemptions_cited`	Which statute(s) (FOIA/OMA) and exemption citation(s) the case addresses.
`public_body` / `requesting_party`	The government body and requester/appellant, as named in the case.
`disposition`	Outcome of the case as decided by the court.
`relevance`	`relevant` or `mention_only` — whether the case decides a FOIA/OMA question outright, or discusses one without it being the central holding. Cases marked `mention_only` carry a note to that effect on their own page.
`summary` / `summary_source`	Case summary and where it came from (model-generated vs. sourced from the opinion's own syllabus, depending on the case).
`best_text_source` / `text_quality`	Which underlying file the extracted case text came from, and a rough quality assessment.
`splice_suspect` / `in_suspect_folder`	Text-integrity flags carried over from the source corpus's own QA process.

Licensing

These are Illinois public records — the underlying Public Access Opinions, determination letters, and court opinions are all already public. This compiled, structured version of that dataset (including the generated summary fields) is offered here for anyone to reuse: for reporting, research, or building your own tools.

The compiled dataset is released underCC0 1.0 (public domain dedication). That covers the parts this project added — the generated summaries, categories, cross-references, and the structure of the export itself. To the extent any copyright exists in that compilation, it is waived. You may copy, adapt, redistribute, and build on it commercially, without asking and without crediting anyone.

The underlying documents were already public records and are not this project's to license either way. Attribution isn't required, but a link back is appreciated and helps readers find the source documents. Note that CC0 waives rights — it makes no warranty that the data is accurate. See Sources & Methods for what has and hasn't been verified.

How to cite this

Illinois FOIA & OMA Reference, 821-document PAC opinions/letters corpus and 287-case caselaw dataset, accessed via illinoisfoia.org. When citing an individual document or case, link its page on this site directly (e.g.illinoisfoia.org/opinions/<id> or illinoisfoia.org/cases/<dedup_key>) so readers can reach the same source text and PDF you did.