Extended Matrix Development Projects

Three layered serializations of stratigraphic physical relationships — pyArchInit `us_table.rapporti` column, GraphML `physical_relationships` per-node packed string (EM 1.6 yEd palette), canonical edges in the s3dgraphy property graph — are now byte-identical when the in-memory graph between them is unmutated. Combined with PR #28 (Postgres backend) and PR #29 (reverse export), EM 1.6 closes the lossless read+write loop between PyArchInit databases and Extended Matrix authoring.

AI tool to extract stratigraphic data from grey literature and PDFs. Absorbs former DP-49 (unified em_data.xlsx supersedes the legacy stratigraphy.xlsx + paradata.xlsx two-file schema).

Grease Pencil annotation for Technical Stratigraphic Units.

New Reused Special Find node (RSF) with red border for reused elements.

Unified Georeferencing panel in EMTools orchestrating shift + EPSG across BlenderGIS and 3D Survey Collection. GeoPositionNode as a passive mirror of scene state for 1.6; active canonical source in 1.7+ when geometries enter the graph.

yEd qualia presets: measurements, properties, color-coded. Property icon and Qualia Importer exist.

Geo-referenced canvas with Geo-node and swimlane-level qualia declared via a dedicated Swimlane Paradata Node Group.

Algorithm for chronological propagation through the graph. BFS-based engine in s3Dgraphy.

Negative SU, Neutral SU, Wall SU hypergraph with spolia. Re-targeted to EM 1.6.

Formalisation of UN (Unità Narrative) and the EMN (Extended Matrix Narratives) extension proposed in Paola Frigerio's master thesis (Padova, A.A. 2022–2023, relatore E. Demetrescu, correlatrice C. Previato). The EMN brings intangible cultural heritage — oral traditions, dialogues, ritual events, social customs — into the EM workflow as graph entities sitting alongside the material reconstruction, with the EM itself acting as the academic side's diario di lavoro that validates the creative output. UN are USV-shaped nodes whose properties combine documentary attestation (registro, contenuto) with creative choice (tono, tempo); they participate in three classifying cases (caso singolo / sistema o scena / racconto dentro al racconto). DP-29 carries the thesis prototype from its theoretical state into the EM 2.0 language formalisation, in coordination with the Universidad de Alicante team that handles character apparenza / movenze / comportamenti.

Boolean + LOD surface generation strategy for Surface Areale. Split out from DP-50 because it is more invasive on source geometry and needs longer API stabilization.

Spreadsheet template (source_list.xlsx) for the systematic registration of bibliographic and archival sources of an EM project.

Territorial graph with topographic units, sites, Canvas concept, and urban stratigraphies (elevation points, core samples).

Import and annotation system for geophysics data.

Formal data transformation steps within the EM workflow.

Subjectivity-specific qualia in EM paradata.

Alternative chronological interpretations as first-class branches in the EM graph. Two complementary inter-branch connectors at the language layer — **excluding** (mutually exclusive, absorbs former DP-42's dashed connector) and **connecting** (concatenation of compatible hypotheses) — plus a downstream EM Tools **Temporal Branch Manager** panel that surfaces the list of alternatives, lets the modeller switch active branch via dropdown, shows a macro description per hypothesis, and filters the US Manager + scene by active branch (parallel affordance to the existing Epoch and Activity filters).

Search engine for graph fragment comparison across studies.

Mapping EM objects to digital twin standards.

Jupyter analytics: GraphML statistics, proxy metrics.

Props and characters with micro-stratigraphy, clickable hypergraphs.

Integration with Heritage BIM and IFC standards.

Node Editor-based graph visualization, Mini Graph for single-SU neighborhoods, and authoring helpers built on top of it.

REST API integration with iDAI.chronontology. Future work targeted at EM 2.0.

Spatialize documents in 3D space, import from Tropy.

Start and End epoch values displayed next to epoch name in swimlanes.

Pipeline for assembling a single target GraphML from multiple input sources (XLSX tables, Blender scene data, manual annotation). Split out from DP-48.

Automate publishing of EM datasets, examples and case studies to the EM Zenodo community, with a custom metadata system that lets EM-aware tools search and retrieve them. Closes the loop between authoring (EMtools / Heriverse) and FAIR data deposit.

Structured documentation form for Virtual Stratigraphic Units.

EM applied to auditory heritage.

EM applied to movement-based intangible heritage.

Virtual SUs driven by cultural patterns, not physical evidence.

Optional ORCID-based identity layer for EMtools that unlocks federated services (Zenodo upload, Heriverse collaborative scenes, AuthorNode/Canvas-header auto-fill) while keeping every authoring feature usable offline / anonymously.

Community-curated registry of EM mapping JSONs (pyArchInit, EMdb, Generic Excel and beyond), served from a dedicated subdomain with per-mapping cards, an embedded web-based builder, one-click download, and a publish-as-PR flow. Closes the loop between the new SQLite/PostgreSQL pyArchInit backend (PR #28) and the long tail of project-specific source schemas the EM community already deals with privately.

EM today writes RDF / CIDOC literals as plain `xsd:string` — no language tag, no SKOS prefLabel/altLabel discipline. DP-63 introduces the RDF 1.1 `@xx` language-tag convention end to end: curated `@it` translations alongside the existing `@en` on `em.ttl`, language-tagged literals from `rdf_exporter.py`, a `lang` field on the s3dgraphy datamodel's text properties, and importer / UI plumbing that detects or assigns the source language. Unlocks proper SPARQL `FILTER(lang(?l) = 'it')`-style queries and aligns the EM ontology + paradata with CIDOC-CRM / SKOS multilingual idioms.

Direct in-viewport authoring of physical stratigraphic relationships. The modeller sees the photogrammetric mesh and the proxies already placed on it (via DP-46 Proxy Box Creator or DP-50 Surface Areale System), clicks on proxy A, drags onto proxy B, picks a relation from a radial pie menu, and the canonical edge lands in s3dgraphy — no context switch to yEd, no detour through the Stratigraphy Manager panel. Temporal relations stay out of scope on purpose: they derive from the physical layer, never the inverse, so the viewport annotator focuses exclusively on what the modeller can directly observe from the photogrammetric evidence.

The graph-level home for ALL graph-scope paradata, held as first-class nodes attached to the graph root: georef shift (GeoPositionNode), author / license / embargo, and any graph-wide document / extractor / property. The third paradata-aggregation tier alongside per-US (DP-60) and swimlane (DP-19).

A property of a contained stratigraphic unit (SF) can be reused on its container (VSF) in two complementary ways. **Pattern 1 — pure inheritance**: the container asserts the same value verbatim, via the new edge `was_derived_from` (mapped to `prov:wasDerivedFrom` in RDF). Authoring shortcut: `material@SF10` on the container's PropertyNode name. **Pattern 2 — premise reuse**: the contained-unit property becomes a direct premise of a Combiner that computes a new value on the container (e.g. column height inferred from fragments' diameters via formula), via the new edge `is_premise_of` (mapped to `crminf:J5_used_as_premise`). Architectural constraint enforced at parse time: an Extractor never accepts a PropertyNode as source — a property is already a mature belief, not raw evidence; the right way to feed it into another node's reasoning is through a Combiner (Pattern 2) or as a pure inheritance pointer (Pattern 1).

A future EM graph editor lets the user switch the active visual clustering register — excavation/location vs activity vs time — rendering one folder register at a time, while the graph keeps all memberships as nodes + edges. Resolves the yEd no-overlapping-folders limitation that makes activity and excavation clusters collide.

Two complementary pieces of work that together separate a real semantic distinction Extended Matrix has been glossing over: **the absence of material can BE a stratigraphic unit, and there are two distinct cases**. The new EM class `NeutralStratigraphicUnit` (node_type `USNV`) captures empty volumes — rooms, doorway openings, evacuated conduits — that have been authored today as filled boxes or workaround US placeholders. The existing `NegativeStratigraphicUnit` (USN, mapped to `crmarchaeo:A3_Stratigraphic_Interface`) stays unchanged and continues to model cut surfaces, erosion lines, contact interfaces — i.e. boundaries with indeterminate vertical extent. On the rendering side, the Visual Manager (DP-35) gains a class-aware dispatcher that maps node class to render mode: USNV → wireframe or low-alpha transparent (toggle, default wireframe), USN → single-sided surface, regular US-family → filled as today. The epoch / chronology / author / license colour carrier (DP-32 propagative resolver) is preserved on the wireframe edges so the chronological reading at a glance survives even when the volume is rendered empty. Project-level override `em_tools.usn_render_mode` (auto / force_filled / force_wireframe) lets the modeller temporarily restore filled rendering for cover renders or sketch-style outputs. Heriverse export carries the same semantics so a published reconstruction shows rooms as voids in the browser too.

Two distinct families of stratigraphic group nodes have been informally conflated under the existing 'anastylosis' name — and a third pattern that doesn't fit either of them keeps surfacing on real fieldwork. DP-68 separates them cleanly. **AnastylosisNodeGroup** is a present-day simulation that assembles real fragments (SF + RSF) in their post-deposition state — it can also play the role of a complex source for a past reconstruction (Anastylosis Study). **RestorationNodeGroup** is a past-state reconstruction that uses virtual stratigraphic units (USV* + VSF) plus surviving real elements. The third pattern is **new-from-virtual**: a present-day physical build that completes or instantiates a past Virtual SU, linked back to the source USV via the `was_derived_from` edge from DP-66 — the Tommaso & Sara case in Turkey. The discrimination is principled, not cosmetic: VSF inside a group flips its temporal anchor from present to past, so a single class doesn't cover both.

s3Dgraphy to triple store / knowledge graph connector and database backend. RDF/TTL and Neo4j export planned.

Case study: equality and same-epoch connectors.

Formalize the DocumentNode's capacity to wrap non-traditional entities (US, subgraphs, external graphs) for comparative reuse and cumulative reconstruction.