Core Concepts #

SemSpect presents graph data in an aggregated schema-oriented manner. It replaces dense instance-level node-edge diagrams with a schema-level group-connection paradigm, enabling the exploration of millions of nodes and their relationships.

The Group Paradigm #

The fundamental element of the exploration is the group.

Instead of rendering 1,000 distinct nodes, SemSpect renders a single group captioned with the label name and the number of nodes.

• Instance View: Small groups display individual dots (e.g., “Product (77)”).
• Aggregate View: Large groups display the total count (e.g., “Order (830)”).

The transition between these views is controlled by the Max. nodes Threshold.

Visualization Logic

• Groups: A disk represents a set of nodes.
• Connections: A line represents a set of relationships between nodes of the groups.

The Exploration Loop #

SemSpect functions as an iterative query builder. The workflow follows a four-step cycle:

1. Seed (Start) #

Populate the canvas by dragging a label from the Label Tree sidebar. Alternatively, use Global Search to find a specific node and start exploring from that individual.

• Example: Drag the label Product to the canvas.

2. Expand (Traverse) #

Reveal connections by expanding a group along a specific path. Select a group to view outgoing relationship types and choose a target label.

• Action: Select Product and expand via includesProduct to Order.
• Result: A new group (Order) appears, containing only the orders that include the selected products.

3. Refine (Filter) #

Narrow down large groups using Filters and Histograms.

• Action: Filter the Order group by orderDate to keep only “2023” values.
• Result: The group count updates (e.g., from 830 to 45). Subsequent expansions will only traverse this filtered subset.

4. Inspect (Drill Down) #

View raw data for verification. Select any group to open the Node Table to inspect specific properties, or view the Connection Table to see the connections between groups.

Inputs & Outputs #

Data entry and extraction points:

Inputs #

• Label Tree: Seed explorations via Drag & Drop.
• Saved Exploration: Restore a previous session state.
• Custom label: Load a pre-defined, complex data segment.
• Deep Links: Initialize via a shared URL or JSON parameter.

Outputs #

• Visual Exports: Export the canvas as SVG or PNG images.
• Data Exports: CSV download of the Node Table or Connection Table.
• Query Snippets: Copy the generated Cypher script for external use.
• Custom Labels: Persist complex filter sets as reusable definitions.

Operational Scope #

Schema Flow vs. Network Topology

The canvas visualizes the logical exploration path, not a static network map.

Because the graph is built iteratively:

1. Connections represent the flow of the query (Source → Target).
2. An individual node may appear in multiple groups simultaneously if it satisfies the criteria of multiple exploration steps.