# lexicon.financial

> Open reference of financial, business, and legal terminology. 28,905 canonical terms across 18 source corpora, plus 1,718 supplementary plain-language definitions. Every entry carries its own license; source attribution is rendered on every page. **Authoritative-source-first** — where FASB, FIBO, or the IRC defines a concept, the lexicon defers to that IRI; the curated `lex:` extension only covers M&A-specific concepts (Quality of Earnings, SDE, DSCR, SBA programs, LOI/NDA/CIM, Earnout, Holdback, MAC) and Dealright-internal vocabulary.

## What this is for

LLMs and agents working on accounting, M&A, lending, tax, or business-process tasks need a reliable referent for terminology — what does *Quality of Earnings* mean exactly, which IRI is the canonical `Goodwill`, what is the difference between `SBA Eligible` and `SBA Prequalified`. This site exposes that referent as static HTML pages plus a JSON API, an MCP server, and a SPARQL-style `sql_query` tool against the underlying Postgres.

## Corpora

- **gaap** (17,182 terms) — XBRL US-GAAP 2026 taxonomy from FASB — every line-item element ships in the U.S. financial-reporting standard. FAF royalty-free.
- **fibo** (3,330 terms) — Financial Industry Business Ontology (EDM Council). Banking, securities, debt, corporate structure. MIT.
- **naics** (2,122 terms) — NAICS 2022 industry codes — full Census hierarchy through 6-digit. US public domain.
- **tryton** (1,201 terms) — Tryton account_us — most US-GAAP-native OSS account hierarchy. GPLv3+ paraphrased.
- **tax** (1,143 terms) — U.S. Code Title 26 (Internal Revenue Code) — 'the term X means Y' definitions extracted from the USLM XML feed. US public domain.
- **business** (1,007 terms) — schema.org business vocabulary (Organization, Product, LocalBusiness, Offer, etc.). CC-BY-SA 3.0.
- **wikidata** (752 terms) — Wikidata finance subtree (~750 classes under Q43015 finance + Q650241 financial institution). CC0.
- **fred** (669 terms) — St. Louis Fed FRED API category taxonomy (top 3 levels). US public domain.
- **bls** (580 terms) — Bureau of Labor Statistics glossary — labor-market vocabulary (employment, unemployment, payroll, occupational classifications). US public domain.
- **legal** (255 terms) — LKIF-Core legal ontology (roles, norms, actions, processes) from EU ESTRELLA. CC-BY 4.0.
- **bea** (200 terms) — Bureau of Economic Analysis glossary — national-accounts vocabulary. US public domain.
- **plaid** (120 terms) — Plaid Personal Finance Categories — bank-feed transaction taxonomy (16 primary + 104 detail). Published spec.
- **gnucash** (100 terms) — GnuCash chart-of-accounts (US common + business templates). Names + types as facts; ~30 paraphrased definitions added. MIT.
- **erpnext** (95 terms) — ERPNext standard chart-of-accounts (account_number ranges + categories). GPLv3 paraphrased.
- **ext** (62 terms) — Curated Lexicon M&A extension (Quality of Earnings, SDE, LOI, working capital, etc.) plus cross-reference anchor concepts (Chapter 7/11/13, FIFO, GAAP, IFRS, etc.). MIT.
- **prov** (49 terms) — W3C PROV-O provenance ontology — parent classes for the dealright Person / Dossier / Evidence / Collector hierarchy. W3C Document License.
- **irs** (21 terms) — Hand-triaged definitions from IRS Pubs 334 / 463 / 583 (trade or business, sole proprietor, cash method, accrual method, etc.). US public domain.
- **ofx** (17 terms) — OFX 2.x TRNTYPE bank-transaction-type enumeration. Spec facts.

## How to query

### MCP server (preferred for agent integration)

Streamable-HTTP JSON-RPC at `https://lexicon.financial/mcp`. Add to a Claude Code session with:

```sh
claude mcp add --transport http lexicon https://lexicon.financial/mcp
```

Six tools, all read-only:

- **`lookup_term({ query, sources? })`** — substring lookup over labels (~20 hits). Optional `sources` array narrows to specific corpora.
- **`search({ query, sources? })`** — hybrid FTS + Voyage AI cosine rerank, source-priority weighted (GAAP wins on accounting basics).
- **`get_term({ iri })`** — full row by IRI, including license + source URL.
- **`get_neighbors({ iri })`** — formal ontology edges (rdfs:subClassOf, owl:equivalentClass, skos:broader/narrower/related).
- **`backlinks({ iri })`** — reverse direction; everything that points AT this IRI.
- **`find_similar({ iri, k })`** — pure cosine-similarity over Voyage embeddings.
- **`compare_terms({ a, b })`** — cosine similarity + lowest common ancestors + direct-edge check between two IRIs.
- **`lookup_by_db_column({ table, column, app? })`** — given a column in an app database, return the lexicon term it's mapped to.
- **`sql_query({ sql, params? })`** — read-only SELECT against the lexicon_* tables.

### REST API

Same endpoints in URL form, JSON output:

- `GET https://lexicon.financial/lookup?q=<substring>`
- `GET https://lexicon.financial/search?q=<phrase>`
- `GET https://lexicon.financial/term/<URL-encoded-IRI>`

### Per-term machine-readable formats

Each term page also serves:

- `<page>.jsonld` — schema.org/DefinedTerm JSON-LD (canonical, citable)
- `<page>.nt` — N-Triples triples for ontology consumers

## URL structure

- `/financial/fibo/{module}/{Class}` — FIBO classes
- `/financial/ext/{Name}` — Lexicon extension
- `/financial/ext/dealright/{Name}` — Dealright-specific dossier vocabulary
- `/financial/ext/concept/{Name}` — Cross-reference anchor concepts (Chapter 7, FIFO, GAAP, etc.)
- `/financial/gaap/{Element}` — XBRL US-GAAP 2026 elements
- `/financial/{plaid|coa|tryton|erpnext|ofx|wikidata}/{Name}`
- `/business/{schema.org Type}` — schema.org
- `/business/{naics|bea|fred}/{Code}`
- `/legal/tax/26/...` — U.S. Code Title 26 (Internal Revenue Code)
- `/legal/tax/irs/{Term}` — IRS Pubs
- `/legal/{lkif-module}/{Class}` — LKIF-Core legal vocabulary

Full inventory at `https://lexicon.financial/sitemap.xml`.

## Source-priority for cross-source label collisions

When the same label resolves to multiple corpora, the autolinker and search ranker prefer (lower number wins):

1. GAAP — XBRL US-GAAP 2026 (the U.S. accounting standard)
2. Lexicon ext / Dealright — curated M&A vocabulary GAAP doesn't cover
3. FIBO — formal financial-industry ontology
4. Concept anchors — cross-reference targets (Chapter 7, FIFO, etc.)
5. Tax / IRS / legal — statutory and agency definitions
6. schema.org / Wikidata — general business / world knowledge
7. Plaid / NAICS / BEA / FRED / OFX — taxonomies and categorizations
8. GnuCash / Tryton / ERPNext — open-source chart-of-accounts (heavily demoted because ~95% of their labels are plain-English variants of GAAP concepts)

## Licensing

Pages embedding CC-BY-SA content (Wikipedia / Wiktionary / schema.org / LKIF) are themselves CC-BY-SA 4.0 (ShareAlike). Pages with only MIT / CC-BY / CC0 / US-public-domain blocks inherit the most-restrictive license — typically MIT for pure FIBO pages. Per-block attribution is rendered on every page.

## Repository

[github.com/dealright/lexicon-financial](https://github.com/dealright/lexicon-financial) — loaders, schema migrations, and the Astro/Vercel/Supabase deployment.