Skip to main content

Why a Graph Model?

Archaeological data is deeply relational, forming an inherently interconnected web of things, places, events, processes, actors, and meaning. These relationships are not simple one-to-one or hierarchical structures; they are complex, many-to-many, and often dynamic.

A relational model (the kind used in spreadsheets or SQL databases) forces such complexity into rigid table structures: rows, columns, and foreign keys. While this approach works well for standardized data—such as inventories or accounting records—it struggles with the rich, overlapping networks that characterize archaeological and cultural research. For example:

  • A single artifact may be connected to multiple actors (who made it, who used it, who discarded it, who recovered it, who identified and analyzed it, who published it, who has current possession and responsibility for it, etc.), places (where it was found, where it was produced, where its raw materials were sourced from, where it was analyzed, and where it is currently stored or on display, etc.), and periods (when it was produced, when it was altered, when it was discarded, when it was recovered, and when it was restored, placed on exhibit, etc.). It may also simultaneously belong to multiple categories (material composition, functional use, etc.).
  • ADDITIONAL POINT.
  • ANOTHER POINT.

Representing this kind of semantic and interpretive richness in a relational model would require complex join tables, nested queries, and repetitive data duplication—quickly becoming difficult to query, visualize, and maintain.

By contrast, in a property-graph model such as Neo4j's:

  • Each entity (a node) directly connects to the entities it relates to via relationships, stored as first-class data.
  • Both nodes and relationships can have properties (key–value pairs) that describe their characteristics—e.g., GIVE EXAMPLE HERE.
  • Traversing the network is intuitive: queries follow natural connections ("find all EXAMPLE HERE").

The resulting structure is both machine-readable and visually interpretable, supporting analysis, visualization, and narrative insight.