Bridges between vocabularies
No vocabulary covers everything. This page maps the five ways the corpus connects vocabularies to each other and to outside knowledge — from declared edition correspondence to hand-written mapping packages — with a note on how mature each one is.
No commands are needed to follow this page. Each claim carries a status in words with a letter mark; the full legend is on the topic index.
Edition alignment
Section titled “Edition alignment”An align block declares correspondence between two editions of a
source, with a relation, a status, and a reviewer. Four maturity
facts, kept in sync with the alignment notes:
- Declaration support: the grammar compiles in real packages (S).
- Reading tools: no structural query returns align rows — the links question covers produces, reads, anchored, alias, and subtype only, so do not build tooling that assumes align rows are queryable (S).
- Adoption: broad — over two thousand blocks across dozens of source files observed in the corpus (S).
- Semantic effect: none — the block stays document metadata and never becomes rule content (S).
Hand-written mapping packages
Section titled “Hand-written mapping packages”Where alignment exists today, it is an ordinary package of hand-written mapping rules, not generated output — a team convention (T). The precedent maps shared dictionary concepts — places, currencies, countries, courts — to external knowledge-base records. Prefer this pattern when the mapping needs human judgment: it is versioned, pinned, and reviewed like any other package.
Concept catalog
Section titled “Concept catalog”A derived full-text index over compiled packages helps authors find candidate concepts before they coin new ones; it requires a source checkout (I). The catalog never merges semantics — it suggests, and the author still writes an explicit import or mapping. Its command surface offers build, search, candidates, and show operations over the compiled snapshot.
Narrow alignment profiles
Section titled “Narrow alignment profiles”A dispatcher runs bespoke, hard-registered bridge modules for narrow pairs: physical quantities against a units knowledge base, and order theory against two proof assistants — experimental (X). Each bridge is custom code for its pair, with its own policy and report schemas. This is the right shape when two formal systems need a precise, verified seam — and the wrong shape for anything broad, where the bespoke cost never pays off.
“Verified” here has a hard boundary: the check runs under a declared
interpretation, and semantic equivalence is not proved — the same line
the import report format draws with
declared_mapping_assumption and semantic_equivalence_proved: false (see the alignment
boundary).
Import mapping
Section titled “Import mapping”External knowledge arrives through versioned import recipes: a recipe
pins its inputs and its mapping, and the ingest surface builds or
checks it (law ingest check <recipe.json>). This requires a source
checkout (I): ingest is absent from the public composition; see
Import.
Mapping profiles
tune how much of the external graph becomes rules versus plain
retrievable knowledge. What does not fit stays out loudly: profile
reports carry a per-decision status and only accepted entries
proceed, and a mismatched recipe is refused with an error rather than
coerced silently (I).
A newer source-to-facts mapping contract exists as a schema with no confirmed engine surface yet — schema only (O).
Which bridge to reach for
Section titled “Which bridge to reach for”- Same concept needed by two acts → share one dictionary import; no bridge at all. Counterexample: mapping between two local copies of “country” instead of importing the shared dictionary doubles maintenance and silently forks meaning.
- Correspondence between two existing vocabularies → a hand-written mapping package (T). Counterexample: a type alias cannot join two dictionaries — it only renames one type.
- Correspondence between two editions of one source → an
alignblock. Counterexample: two acts in different languages are not two editions of one source; aligning them asserts a correspondence the sources never made. - Candidate discovery before coining anything → the concept catalog (I). Counterexample: a catalog hit is a pointer, never an identity claim — it never licenses skipping the explicit import.
What bridges are not
Section titled “What bridges are not”Some subject-matter packages reuse the word “bridge” in their names for ordinary domain rules — tolling rules, identity rules, forensic linkage. Those are regular packages, not mapping mechanisms, and this page makes no claims about them.
Where to go next
Section titled “Where to go next”- The companion page on local and shared vocabulary covers the dictionaries that mappings usually target.
- The companion page on the audit surface shows how to query who reads what across packages.
- For construct-level detail, see the Vocabulary construct reference.
- For structural queries over links, see LawQL.
Documentation for Arxo. Writings — blog.arxo.io.
Anonymous visit counts on stats.arxo.io, no cookies.