Synthetic Form 1120-POL - U.S. Income Tax Return for Certain Political Organizations Data
Synthetic training data — no real PII, fully coherent identities
Generate synthetic Form 1120-POL returns for political organizations — taxable investment income, deductions, and the specific-deduction computation on one page. Training data for campaign-finance compliance tooling and exempt-organization document pipelines.
63
Fields per document
1
Page
Tax
Category
What this document is
Form 1120-POL is the income tax return filed by political organizations — campaign committees, PACs, and party committees — on their taxable income. Contributions themselves are exempt function income and are not taxed, so the return is concerned with what the organization earns on the money it holds: interest, dividends, rents, and gains, less allowable deductions and the specific deduction available to political organizations. It fits on a single page.
Why generate synthetically
Campaign-finance and exempt-organization compliance is a document-heavy field with essentially no shareable training data, and the 1120-POL sits at an awkward intersection: it looks like a corporate return, files like one, but its income profile is nothing like a business's. Synthetic 1120-POLs give compliance and disclosure platforms a labeled corpus for a form their pipelines otherwise have to handle by hand-written rules.
What makes synthetic data useful
Each return is generated for a coherent entity, with investment income components — interest, dividends, rents, royalties, capital gains — scaled together so that the total income line is a genuine sum rather than an independent draw. Because political organizations typically hold funds rather than operate, most returns show interest as the dominant income line and leave the operating-income categories empty, which is the realistic shape of this filing.
Training challenges
The 1120-POL is visually a member of the 1120 family but semantically unlike its siblings: an extractor that has learned corporate patterns will expect gross receipts and cost of goods and find neither. Most income lines are legitimately zero — median rents, royalties, capital gains, and other income are all zero in our corpus — so the page is dominated by explicit zeros and blanks that a model must not treat as failures. A date field and twelve checkboxes sit among the currency cells on the same sheet, mixing three extraction modes in a small area, and the specific-deduction line is a fixed statutory amount that looks like a data value but is not derived from the organization's activity.
Generate synthetic Form 1120-POL - U.S. Income Tax Return for Certain Political Organizations data
Start with 500 free credits. No credit card required.
Generate NowWho uses this data
Campaign-finance and political-compliance software, exempt-organization tax platforms, nonprofit accounting and filing services, and the regulatory-technology teams that ingest committee filings at scale. It also serves as a hard negative class for corporate-return classifiers: an 1120-POL will pass a shallow 'is this an 1120?' test and then fail every downstream corporate extraction rule.
Document complexity profile
63 fields on a single rendered page: 31 currency amounts, 19 text, 12 checkbox targets, and 1 date. 69 annotation relations — more than the field count, because shared labels govern paired entries. There are no conditional or arithmetic bindings in the definition; the totals are supplied as a consistent set by the generator, and the extraction difficulty is the mix of three field modes at close spacing rather than expression depth.
Key stats from our synthetic corpus
Quantitative characteristics of the Form 1120-POL - U.S. Income Tax Return for Certain Political Organizations documents our generator produces.
| Metric | Value | Detail |
|---|---|---|
| Population coverage | 64% | 64% of identities in our 1,000-identity corpus are attached to an entity and therefore 1120-POL eligible. |
| Median taxable interest | $1,650 | Median interest income is $1,650 with a p75 of $4,419 — the dominant income line on a political organization's return, since the organization earns on held funds rather than operating. |
| Median total income | $5,683 | Median total income is $5,683, p25–p75 of $2,588 to $13,405, maximum $208,287. The long right tail comes from larger committees holding substantial balances. |
| Income lines at zero | 4 of 7 | Median rents, royalties, capital gains, and other income are all zero. Four of the seven income categories are empty on a typical return, which is the single most important thing to train into a model reading this form. |
| Annotation relations | 69 | 69 label-to-value relations on a page carrying fewer fields, because printed labels govern paired entries. Relation supervision is what disambiguates adjacent short labels at tight line spacing. |
How this document co-occurs with others
Rates at which identities in our corpus that produce a Form 1120-POL - U.S. Income Tax Return for Certain Political Organizations also produce other documents.
| Correlation | Rate | Detail |
|---|---|---|
| Other niche 1120 variant (1120-H) | 100% | Form 1120-H is the other compact single-page 1120 variant. The two are the canonical confusion pair for entity-type classification within the 1120 family. |
| Standard corporate return | 100% | The same entity can also produce a standard Form 1120, giving a corporate/political contrast pair over a shared visual family — the hard negative that shallow classifiers fail. |
| Can file a 7004 extension | 100% | The same entity can produce a Form 7004 extension, whose form code identifies which return is being extended. |
| Files quarterly payroll returns | 100% | Committees with paid staff file Form 941 quarterly, and every entity in our corpus does, so organization tax and payroll filings generate together. |
| Entity W-9 on file | 100% | Every entity can co-generate a W-9 carrying the organization name and TIN on the return header — the identity anchor for organization-resolution training. |
All stats above are corpus-derived: they were computed on a local synthetic corpus of 1,000 generated identities produced by SymageDocs' World Simulation Engine, with 1120-POL figures measured over the entity-eligible subset. No real political organization, committee, or donor data was used. `make corpus-stats` regenerates the shipped corpus_stats.json for the site's analytics form set, which does not include this edition; reproduce these figures with `python -m symagedocs.corpus_analysis --corpus-size 1000 --form-id irs_f1120pol_standard_2022`.
Frequently asked questions
- What data format do synthetic Form 1120-POL documents include?
- Each generated identity produces a rendered PDF plus a structured JSON annotation file with bounding boxes and ground-truth values for all 63 fields on the single page. COCO, YOLO, FUNSD, and BIO/NER exports come from the same job.
- Why are most income lines zero?
- Because that is what a real political organization's return looks like. Median rents, royalties, capital gains, and other income are all zero in our corpus; interest is the dominant line, with a median of $1,650. Training on a corpus where most numeric lines are legitimately zero is the only way a model learns not to hallucinate values into empty cells.
- Does the form include contribution amounts?
- No, and neither does the real form. Contributions are exempt function income and fall outside the 1120-POL's taxable computation. If you need contribution-level data, that lives in campaign-finance disclosure filings, not in this return — a distinction a document-classification model has to encode.
- How does labeling work?
- Every value field and every printed label is emitted with its exact bounding box and label-to-value relation, including the checkbox targets and the date field, so a model can be supervised on mixed-mode extraction within a single dense page.
- Can I use this data commercially?
- Yes. All synthetic data is generated from statistical models, contains no real organization, donor, or committee data, and is licensed for commercial use including ML model training and benchmarking.