Early developmentNot everything described here has shipped yet.

NeoWiki
Get a demo

Graph Model ​

NeoWiki stores a query-optimized projection of its data in a Neo4j graph database (ADR 3). The source of truth for all data remains in MediaWiki revision slots (ADR 4); the graph is a secondary store that enables efficient querying and relationship traversal.

For definitions of domain terms like Subject, Statement, and Schema, see the Glossary.

Overview ​

The graph consists of two node types and two relationship categories:

(:Page)-[:HasSubject {isMain}]->(:Subject:SchemaName)
(:Subject)-[:RelationType {id, ...}]->(:Subject)

Page nodes represent MediaWiki pages and carry page-level metadata. Subject nodes represent structured data entities. HasSubject relationships connect pages to their Subjects. Typed relationships connect Subjects to other Subjects via Relations.

Page Nodes ​

Every page that contains structured data has a corresponding :Page node. These nodes make page metadata available for graph queries.

Page identity is scoped per wiki. A shared graph can hold pages from multiple wikis (e.g. a wiki farm), and MediaWiki page ids are only unique within a single wiki. Each page node therefore carries a wiki_id and is identified by the (wiki_id, id) pair (ADR 22). In a single-wiki install there is one wiki_id value and behaviour is unchanged.

PropertyNeo4j TypeDescription
wiki_idstringMediaWiki Wiki ID (database name + table prefix) of the owning wiki
idintegerMediaWiki page ID (unique per wiki)
namestringFull page title, including the namespace prefix (e.g. Help:Installation)
namespaceIdintegerMediaWiki namespace ID of the page (e.g. 0 for the main namespace, 12 for Help)
creationTimedatetimeWhen the page was created
lastUpdateddatetimeWhen the page was last modified
lastEditorstringUsername of the last editor
categoriesstring[]MediaWiki categories the page belongs to

creationTime and lastUpdated are stored as Neo4j datetime values (ISO 8601), converted from MediaWiki's YmdHis timestamp format.

namespaceId holds MediaWiki's canonical namespace ID. Built-in namespaces (e.g. 0 main, 12 Help, 14 Category) have the same ID on every wiki, so filtering by namespaceId behaves consistently across a graph shared by multiple wikis. Custom namespaces defined via $wgExtraNamespaces get per-wiki IDs that may differ in meaning between wikis, so pair namespaceId with wiki_id to filter those unambiguously.

Subject Nodes ​

Each Subject stored on a page gets a :Subject node. Subject nodes carry two labels: Subject and the name of their Schema (e.g., :Subject:Person, :Subject:Company). The Schema label changes if a Subject's type changes.

Fixed properties ​

PropertyNeo4j TypeDescription
idstringSubject ID, 15 characters starting with s (unique)
namestringSubject label
wiki_idstringMediaWiki Wiki ID of the wiki that owns the Subject

Unlike page ids, Subject ids are globally unique nanoids (ADR 14), so a Subject's identity is its id alone. The wiki_id is carried only for per-wiki query filtering in a shared graph; Subject-id namespacing is deferred (ADR 22).

Dynamic properties ​

Each Statement on a Subject becomes a node property, keyed by Property Name. The value is converted to a Neo4j-compatible format by the corresponding PropertyType implementation. For example, a Statement with Property Name "Founded at" and a number value of 2019 results in a node property Founded at: 2019.

Relation-type Statements are not stored as node properties. They are stored as relationships between Subject nodes (see below).

Relationships ​

HasSubject ​

Connects a Page node to each of its Subject nodes.

PropertyNeo4j TypeDescription
isMainbooleantrue for the Main Subject, false for Child Subjects

A page can have at most one Main Subject and any number of Child Subjects (ADR 7).

Typed Relations ​

Subject-to-Subject relationships represent Relations. The relationship type in Neo4j is the Relation Type defined in the Property Definition (e.g., Has author, Has product). Names that are not valid Cypher identifiers are backtick-escaped.

PropertyNeo4j TypeDescription
idstringRelation ID, 15 characters starting with r
(additional)scalarAny properties from the Relation's property map

When a Subject is deleted but still has incoming relations from other Subjects, its outgoing relationships and HasSubject relationship are removed, but the node itself is kept so that the incoming references remain valid.

Constraints ​

Two uniqueness constraints are intended for the graph — the (wiki_id, id) pair is unique on :Page nodes (ADR 22), and Subject.id is unique. These are not created automatically yet (#874).