Corpus statistics

Live counts across the 18-corpus terminology stack backing lexicon.financial. Recomputed at every deploy from the Postgres source of truth.

28,905canonical terms

21,389with prose definitions

28,843Voyage AI embeddings (1024-dim)

32,849labels (pref + alt)

10,687ontology relations

1,718plain-language supplements

24declared sources

0app DB mappings

Per-source breakdown

Source	Terms	With definitions	Alt labels	Supplements	License
XBRL US-GAAP Taxonomy `gaap`	17,182	14,760 (86%)	1,701	0	FAF royalty-free
FIBO `fibo`	3,330	2,662 (80%)	95	1,714	MIT
NAICS 2022 `naics`	2,122	0 (0%)	2,125	0	US public domain
Tryton account_us (US-GAAP CoA) `tryton`	1,201	0 (0%)	15	0	GPLv3+ (names + types treated as facts; descriptions paraphrased)
U.S. Code (Title 26) `tax`	1,143	1,143 (100%)	94	0	US public domain
schema.org `business`	1,007	933 (93%)	51	4	CC-BY-SA 3.0
Wikidata `wikidata`	752	618 (82%)	8	0	CC0 1.0
St. Louis Fed FRED Glossary `fred`	669	0 (0%)	31	0	US public domain
Bureau of Labor Statistics Glossary `bls`	580	580 (100%)	41	0	US public domain
LKIF-Core `legal`	255	199 (78%)	0	0	CC-BY 4.0
Bureau of Economic Analysis Glossary `bea`	200	200 (100%)	0	0	US public domain
Plaid PFC Taxonomy `plaid`	120	104 (87%)	239	0	Published spec
GnuCash US chart templates `gnucash`	100	68 (68%)	279	0	GPLv2+ (names treated as facts; definitions paraphrased)
ERPNext standard charts `erpnext`	95	0 (0%)	200	0	GPLv3 (paraphrased)
Lexicon extension `ext`	62	62 (100%)	129	0	MIT
W3C PROV-O `prov`	49	22 (45%)	14	0	W3C Document License
IRS Publications `irs`	21	21 (100%)	0	0	US public domain
OFX 2.x Specification `ofx`	17	17 (100%)	30	0	Published spec

Documented but not ingested

Stub source rows for corpora the loader documents but has not ingested. Each has a loader/load_<key>.py stub recording the blocker (deprecated URL, license incompatibility, bot-blocking) and a resume path.

CFPB Consumer Financial Glossary cfpb
Princeton WordNet (financial sense subset) wordnet
Financial Report Ontology (Hoffman) fro
FFIEC Call Report Taxonomy ffiec
Apache OFBiz Accounting Seed ofbiz
FDIC Money Smart fdic

Relation-predicate breakdown

Predicate	Edges
`rdfs:subClassOf`	10,524
`owl:equivalentClass`	163

Most-referenced terms

Terms with the most inbound rdfs:subClassOf / owl:equivalentClass / skos:broader edges. Useful as anchors when reasoning about the graph structure — these are the conceptual hubs.

Countries — 243 inbound edges Countries
registered security — 221 inbound edges RegisteredSecurity
bank — 159 inbound edges Q22687
preferred share with fixed rate dividend — 112 inbound edges PreferredShareWithFixedRateDividend
extendable preferred share — 84 inbound edges ExtendablePreferredShare
exchangeable preferred share — 84 inbound edges ExchangeablePreferredShare
CreativeWork — 74 inbound edges CreativeWork
non-cumulative preferred share — 72 inbound edges NonCumulativePreferredShare
savings bank — 68 inbound edges Q157963
Intangible — 63 inbound edges Intangible

Source	License	Entries	Why we use it
FIBO	MIT	3,330 classes	Canonical financial ontology from the EDM Council — banking, securities, debt, corporate structure.
Lexicon extension (`lex:` / `dr:`)	MIT	62 classes	Curated M&A terms (Quality of Earnings, SDE, LOI, DSCR, BBS grade, Earnest Money, Purchase Agreement, Working Capital, Customer Concentration), Person / Dossier / Evidence / Collector (PROV-O subclasses), and ~30 cross-reference anchor concepts (Chapter 7/11/13, FIFO, GAAP, IFRS, FASB, PCAOB, etc.).
XBRL US-GAAP 2026	FAF royalty-free	17,182 elements (14,760 with documentation)	Every element FASB ships in its 2026 financial-reporting taxonomy — line-item names, abstract groupings, axes, members, types — pulled from `xbrl.fasb.org/us-gaap/2026/us-gaap-2026.zip`. Standard labels become pref labels; documentation linkbase prose becomes definitions; terse / total / period-start / period-end labels join as alt labels.
Plaid PFC Taxonomy	Published spec	120 categories (16 primary + 104 detail)	Plaid's transaction-categorization taxonomy. `INCOME_WAGES` survives as alt-label so downstream Plaid integrations can lookup canonically; pref label is the humanized leaf ("Wages"); detail rows are `rdfs:subClassOf` their primary.
GnuCash chart of accounts	MIT (paraphrased)	100 accounts (13 with definitions)	Account names + types from the GnuCash US chart templates (common + business). Names + type enums treated as facts; the GPL-tainted `act:description` strings are dropped and replaced with hand-paraphrased one-line definitions for ~30 well-known accounts (Cash, Accounts Receivable, Cost of Goods Sold, Depreciation, Retained Earnings, etc.) at `data/coa_definitions.jsonl`.
Wikidata — finance classes	CC0 1.0	752 classes	Pulled via SPARQL from `wdt:P279*` under `wd:Q43015` (finance) + `wd:Q650241` (financial institution). CC0 — no attribution burden.
schema.org	CC-BY-SA 3.0	1,007 classes	General business concepts (Organization, Product, LocalBusiness, Offer) — used across the open web, matches common consumer vocabulary.
NAICS 2022	US public domain	2,122 codes	The full Census NAICS hierarchy, sector through 6-digit industry — same codes brokers see on Schedule C. Each code links `rdfs:subClassOf` the level above; the numeric code doubles as an alt-label so `541211` matches "Offices of Certified Public Accountants".
Tryton account_us	GPLv3+ (paraphrased)	1,201 entries	The most US-GAAP-native account hierarchy in the OSS ecosystem — balance-sheet vs income-statement classification baked into account types. Names + types ingested as facts; no GPL-tainted prose mirrored.
BEA Glossary	US public domain	200 entries	Bureau of Economic Analysis terminology — national-accounts vocabulary (GDP components, NIPA structure, BEA-specific aggregates). Federal complement to BLS labor-market vocabulary.
BLS Glossary	US public domain	580 entries	Bureau of Labor Statistics glossary — labor-market terminology (employment, unemployment, payroll, wages, occupational classifications). Sourced via Wayback Machine snapshot since BLS Akamai infrastructure 403s direct fetches.
Open English WordNet 2024	CC-BY 4.0	578 alt-labels (synonym layer)	WordNet's finance-domain synsets layered as `skos:altLabel` rows on the existing corpus — adds e.g. "AR" / "trade receivables" alts alongside the canonical "Accounts Receivable". Filter: lexicographer file in `noun.act` / `noun.attribute` / `noun.cognition` / `noun.communication` / `noun.event` / `noun.group` / `noun.possession` / `noun.quantity` / `noun.relation` / `noun.state` AND definition matches a finance keyword regex.
FRED Categories	US public domain	669 categories	St. Louis Fed FRED API category taxonomy, top 3 levels — 8 top-level economic domains (Money/Banking/Finance, National Accounts, Prices, etc.) and their sub-categories down to specific measurement areas. Each category links `rdfs:subClassOf` its parent.
ERPNext charts	GPLv3 (paraphrased)	95 entries	Adds `account_number` ranges (1xxx Asset, 2xxx Liability, 4xxx Income, 5xxx Expense) and `account_category` labels that GnuCash and Tryton don't expose explicitly.
OFX 2.x TRNTYPE	Spec (facts)	17 enum values	The bank-statement transaction-type enumeration every U.S. bank-feed downloader uses — CREDIT / DEBIT / INT / DIV / FEE / DEP / ATM / POS / XFER / CHECK / PAYMENT / DIRECTDEP / DIRECTDEBIT / etc. Underneath Plaid's consumer categorization layer.
LKIF-Core	CC-BY 4.0	255 classes	Core legal concepts (roles, norms, actions, processes) from the EU ESTRELLA project — the most-cited open legal ontology.
U.S. Code Title 26 (Internal Revenue Code)	US public domain	1,143 definitions	Explicit "the term X means Y" definitions extracted from the USLM XML feed at uscode.house.gov — §7701 General Definitions plus 1,000+ scattered in-line definitions. Source-linked to Cornell LII.
IRS Publications 334 / 463 / 583	US public domain	21 definitions (hand-triaged from 75 candidates)	SMB-relevant terms inline-defined in the published HTML of IRS Pubs (Trade or business, Sole proprietor, Cash method, Accrual method, Section 179 deduction, Per diem rates, etc.). Triage rules + re-run instructions at `data/irs_triage_notes.md`.
W3C PROV-O	W3C Document License	49 classes (parents only)	Provenance ontology — used as the parent class hierarchy for the dealright-ext Person / Dossier / Evidence / Collector classes. No own page bucket; rendered as parent classes inside dealright-ext term pages.
Wikipedia — supplements	CC-BY-SA 3.0	586 hits	Encyclopedia-length summaries via the REST `/page/summary/` endpoint for richer context on complex concepts (CDOs, derivatives, etc.).
Wikidata — supplements	CC0 1.0	724 hits	One-line structured descriptions attached as an additional block on FIBO pages.
Wiktionary	CC-BY-SA 4.0	322 hits	Plain-language dictionary senses, appended as a supplement where FIBO's formal definition is too terse.
Investor.gov · CFTC · Federal Reserve	US public domain	1,205 entries (82 hits)	Authoritative US regulatory + banking glossaries scraped once into JSONL. Public domain — zero licensing friction.