Data Funnel Structure

The Extended Matrix uses a hierarchical data structure — the Data Funnel — to spread context across the graph without duplicating it on every node. Three scopes hold values at increasing levels of specificity: the Canvas (global default), the Epoch swimlane (period-wide), and the individual stratigraphic node (per-unit). When the resolver looks up a property, it walks these scopes from the most specific to the most general and returns the first non-null value it finds.

┌─────────────────────────────────────────────────────────┐
│   Resolution order — first non-null value wins          │
└─────────────────────────────────────────────────────────┘

   ┌───────────────────────────────┐    ◄── HIGHEST priority
   │  Specific Node                │
   │  (US, USV, USD, SF, ...)      │
   │  start · end · qualia         │
   └───────────────┬───────────────┘
                   │  if not on the node ↓
                   ▼
   ┌───────────────────────────────┐
   │  Epoch Node   (swimlane)      │
   │  start · end                  │
   └───────────────┬───────────────┘
                   │  if not on the epoch ↓
                   ▼
   ┌───────────────────────────────┐    ◄── LOWEST priority (default)
   │  Canvas   (graph header)      │
   │  EM ID · ORCID · License ·    │
   │  Embargo                      │
   └───────────────────────────────┘

In plain prose: a value declared on a single node wins over the value carried by its epoch swimlane; the epoch’s value in turn wins over the canvas-level default. In the opposite direction this is inheritance — a node with no explicit value picks up its epoch’s value, and an epoch with no explicit value picks up the canvas default.

The next three sections describe what each scope typically holds, listed from the broadest (canvas) to the narrowest (specific node).

1. General Background Data

Canvas-level scope — lowest priority, the default fallback that applies to every node when no narrower scope declares the value.

General Background Data encompasses information that applies uniformly to all nodes within the knowledge graph of the Extended Matrix. These data provide a global context, essential for maintaining coherence across different elements and preventing data duplication.

Values:

  • Extended Matrix ID: An identifier assigned to a coherent stratigraphic portion of an archaeological site or monument. This ID helps avoid duplications and maintains consistency in node identification within the graph.

  • ORCID: Unique identifiers of authors involved in the project, allowing each node to be linked to a specific author in a structured and identifiable manner.

  • Licence: It uses the Creative Commons license schema (i.e., CC-BY-NC).

  • Embargo: Expressed in months, it defines the disclosure moment in time when the dataset can be released to the public (using the license above).

To ensure that every node in the graph is coherently linked to a set of common properties, maintaining the integrity of the system and providing adequate granularity.

The Extended Matrix ID serves as a prefix to make each stratigraphic unit identifier unique across multiple excavation projects. This approach solves a common archaeological data management challenge: the same identifier (like “USM100”) may be used across different excavation sites.

By adding a unique excavation ID prefix, we create globally unique identifiers. For example:

GTS16.USM100

Where:

  • GTS16 is the Extended Matrix ID for the Great Temple of Sarmizegetusa, 2016 campaign

  • USM100 is the local stratigraphic unit identifier

Note

Moving from Individual Context to Landscape Analysis

Traditional archaeological documentation manages each excavation as a separate entity, making inter-site analysis challenging. The Extended Matrix ID system allows archaeologists to:

  1. Manage multiple sites in a single Blender session: Combine data from multiple excavations or monuments while maintaining distinct identification.

  2. Prevent ID collisions: Even when two different sites have identically numbered units (e.g., USM100), the prefixed IDs (GTS16.USM100 vs. PTR19.USM100) remain distinct.

  3. Enable landscape-level analysis: By integrating multiple sites with unique identifiers, researchers can shift from analyzing individual contexts to understanding entire archaeological landscapes.

  4. Facilitate collaboration: Team members working on different sites can share a single data environment while maintaining organizational clarity.

This approach creates a hierarchical namespace for archaeological data, similar to domain naming systems used in other fields, ensuring that local identifiers remain meaningful within their context while becoming globally unique when combined with the Extended Matrix ID.

2. Local Background Data

Epoch-level scope — wins over the canvas default; applies uniformly to every node placed in the same swimlane unless overridden at the node level.

Definition: Local Background Data are information that apply only to a subset of stratigraphic nodes. These data include properties shared among certain nodes that belong to the same context or chronological period, defined by a shared temporal property.

Values: - start and end: Define the temporal boundaries of each Epoch - graphically represented with a swimlane - (EpochNode), representing a time interval (e.g., Augustan Era: -27 to 14). These dates are determined through macroscopic interpretation of archaeological evidence such as architectural styles, construction techniques, and formal coherence.

Objective: To connect groups of nodes to a common temporal or functional context, simplifying the representation of multiple nodes sharing similar properties within a specific context.

3. Specific Node Data

Node-level scope — highest priority, wins over both the epoch swimlane and the canvas default.

Definition: Specific Node Data represent the unique information that applies to individual stratigraphic units. These data take precedence over Local Background Data and can override shared properties when necessary.

Values:

  • Start Time and End Time: The specific temporal properties of a stratigraphic unit indicating its chronological limits. These data override shared Temporal Deltas.

  • Qualia: Physical Properties like Material, style, dimensions (height, width, length), and state (existing or destroyed) of the stratigraphic unit or subjective properties to express how thigs were percived in the past (meaning, scope, etc…).

Objective: To provide detailed descriptions of each node’s unique characteristics, offering more granular and precise information compared to Local Background Data.

Data Propagation Protocol

The mechanism by which a query for the value of a property on a node walks the Data Funnel. The rule is the same as the resolution order above and applies uniformly to temporal bounds, authorship, licence, embargo and any other propagatable property:

  1. If the specific node carries the value, use it.

  2. Otherwise, fall back to the value declared on the epoch swimlane the node belongs to.

  3. Otherwise, fall back to the canvas-level default.

  4. If none of the three scopes carries the value, the property is undefined at that point in the graph.

Worked example — temporal resolution on a USM node:

  • The node declares start = 20 AD, end = 40 AD → both bounds resolve to the node’s own values; the node is included in a chronological query for the 30–35 AD window.

  • The node declares start = 20 AD and no end → start resolves to the node, end falls back to the epoch swimlane’s end.

  • The node declares no times at all → both start and end resolve to the epoch swimlane’s bounds.

Objective: ensure that values are correctly inherited or overridden along the node → epoch → canvas chain, keeping the graph consistent without forcing every node to redeclare context that is already shared at a broader scope.

Benefits of the Data Funnel Structure

  • Consistency: By defining data at different levels, the Data Funnel ensures that nodes share common properties where appropriate, reducing inconsistencies.

  • Avoidance of Duplication: General and Local Background Data prevent the need to repeat the same information across multiple nodes.

  • Granularity: Specific Node Data allow for detailed descriptions when necessary, providing precise information for individual nodes.

  • Efficient Data Management: The hierarchical structure facilitates easier data management and updates, as changes at higher levels automatically propagate to relevant nodes.

Important Considerations

Note

Setup of a new GrapML file

  1. Set the ID in the upper part of the canvas. A common error when setting up a new GrapML file is to forget to set the Extended Matrix ID in the General Background Data. This omission can lead to confusion and potential ID collisions when integrating data from multiple excavation projects.

  2. Set start and end for each epoch. You need to set the start and end dates for each Epoch in the Local Background Data. This step is crucial for establishing the temporal framework within which the stratigraphic nodes will be contextualized.

Conclusion

The Data Funnel structure in the Extended Matrix provides a robust framework for organizing stratigraphic data. By categorizing information into General Background Data, Local Background Data, and Specific Node Data, the system balances the need for both shared context and detailed specificity, enhancing the overall integrity and usability of the stratigraphic knowledge graph.