Relation Type Mapping

Select Tools > Map Relation Types to open the dialog. Or use the context menu on a relation and select: "Edit Relation Type..."

The Relation Type Mapping dialog allows you to manage how relation names in your graph map to semantic types and directions. It also provides a legend to control the visibility of different relation types on the graph.

Mapping Table

The table on the left defines the properties for each unique relation name in your collection:

• Relation: The unique name of the relation.
• Reversed: The name used for the reverse form of the relation (e.g., "is parent of" reversed is "is child of"). Used for verbalizations.
• Type: The semantic category of the relation (e.g., Generalization, Composition). See the Type Legend for details.
• Direction: Specifies the semantic flow of the relation. This is critical for features like inheritance and hierarchy building.
  • source → target (Default): The source node is the more specialized concept or part. The target node is the general category or the whole. Example: Car --is_a--> Vehicle.
  • target → source (Reversed): The target node is the specialized concept or part. The source node is the general category or the whole. Example: Vehicle --has_type--> Car.

Understanding Semantic Direction

In a graph, edges have a visual direction from source to target. However, the semantic direction identifies which node is the "parent" (generalization) or "whole" (composition).

Termboard uses this setting to determine:

• Inheritance: Properties flow from the "parent" to the "child" based on the semantic direction.
• Hierarchy View: The tree structure is built by following relations from child to parent.
• Semantic Checks: Cycle detection and multi-parent warnings rely on knowing the hierarchy flow.

Auto-Detection

The Auto Fill action tries to guess the correct direction based on common naming patterns:

• target → source is guessed if the name includes words like: subtype, subclass, instance, part, component, member, includes, contains, generalizes, is parent of.
• source → target is the default for most other patterns like: is a, is part of, belongs to, inherits from, specializes.

Table Actions

The toolbar above the table provides quick actions:

• Clear Filters: Resets any active filters in the table headers.
• Add Row: Manually adds a new relation mapping.

Type Legend

The sidebar on the right serves as a legend and visibility controller:

• Visibility Checkboxes: Toggle the checkbox next to a type (e.g., Generalization) to show or hide all relations of that type on the graph.
• Color Indicators: Shows the color assigned to each relation type.

Legend Descriptions

• Generalization: Hierarchy between general and specific concepts (e.g., Car is a Vehicle). Used for Hierarchy View and inheritance.
• Composition: Strong ownership; the part cannot exist without the whole (e.g., Car has Engine).
• Aggregation: Weak ownership; the part can exist independently (e.g., Team contains Players).
• Attributive: Links an entity to its properties (e.g., Car has property Color).
• Interaction: Actions or processes (e.g., Person drives Car).
• Dependency: One entity depends on another (e.g., Order requires Payment).

Bottom Actions

• Auto Fill: Automatically guesses Reversed names and Types based on common patterns.
• Prune: Removes relation mappings that are no longer used in the current graph.
• CSV / Import: Export the mapping to a CSV file or import a previously saved mapping.
• Save: Applies all changes and updates the graph.