Inference
8 pages filed under Decisions › inference. A decision is a choice made for the project, with its context, its options and what it affects.
| Decision | Date | First line | Links |
|---|---|---|---|
| Bounded neighbourhood | 2026-09-12 | Co-occurrence is accumulated per paragraph and never materialised as a full matrix. Only the fifty best neighbours of each node are kept, ranked by count then by identifier. | 50 |
| MinHash for twin resources | 2026-09-13 | The text similarity that helps recognise the forms of one document works on extracted text, never on binary content, and never compares every pair. Each text goes to its comparison form, then to shingles of five words, then to a MinHash signature of 128 functions under a fixed seed; LSH banding, four rows per band, enumerates the candidate pairs, two signatures that share no band being never compared, so that the full matrix is never built. In auto mode the pairs estimated at 0.5 or more have their exact Jaccard index recomputed on the full shingle sets, and the finding gives the share of lines in common; a pair whose word counts differ by more than half is an inclusion rather than a duplicate and its content signal is capped. Resources are sorted by identifier before anything else, so two runs give the same result. | 21 |
| Links are authoritative | 2026-09-12 | A markdown link to another note is the strongest relation the tool knows: confidence 1.00, above anything inferred, and a frontmatter reference comes next at 0.90. What an author wrote is never overturned by what the tool recognised: the relation of a mapped section or of a typed attribute stands, inference only fills what nobody declared, and the page tells a written link from a recognised word by the way it is drawn. A link to a missing file is an error, a link to a document attaches it, a link from a note to itself yields nothing, and several links to one target are one link with several provenances. | 20 |
| Type-driven neighbour order | 2026-09-13 | The neighbourhood panel of a page lists neighbour types in a priority order declared per type in the profile: operations first on an API, accessed objects on a screen, what it applies to on a rule. Every displayed neighbour carries the rank of its type; unlisted types and keyword pages share the last rank, confidence decides within a group, and the panel is truncated after this ordering. A type without a declaration keeps the order by decreasing confidence. The order is profile data, never a condition in a template: the panel renders the list as received and separates the groups by rank alone. | 17 |
| Types as modules | 2026-09-13 | A type of the profile is a folder that carries everything the type needs: its declaration, its labels per interface language, its note template and, when the generic page is not enough, the components that render it. The core types are written that way and the default profile is assembled from them, so that the format the engine ships is the format a team extends it with: a project keeps its own modules in a folder its profile names, a plugin contributes modules through a contribution point of the plugin API, and the build merges them before the keys of the project profile, plugins first, refusing a module of a core type, which is extended through the profile, and two modules of one type. The site never learns a type slug: the entity page exposes the declaration of the type and every attribute of the note, declared and not, and a dedicated component for a type, an attribute or a mapped section is resolved by name, the project theme first, then the module, then the generic page, which remains correct when nothing specific is said. | 17 |
| Declarative profile | 2026-09-12 | The meta-model is a YAML profile validated by a schema. The engine knows only types, attributes, relations and scores. Adding a type or a relation pair is a profile change, never a code change. | 15 |
| Keyword page threshold | 2026-09-12 | A keyword page exists from three occurrences in at least two files; below that, the word is searchable but has no page. The threshold is configurable, and the build reports how many pages it generated and how many expressions it discarded. | 10 |
| Confidence and provenance on every link | 2026-09-12 | Every link carries a confidence between 0 and 1 and at least one provenance. When methods agree, confidences combine and every provenance is kept. The first version stores the score, orders mentions with it, and does not show it. | 9 |
8 inference of 8 — pagination by twenty. The Links column counts the related pages, which brings the most central inference of the journey to the top.