Saved Queries
How to save, load, and share graph explorations across explorer views.
What is a Saved Query?
A saved query is an ordered list of graph operations — each tagged as additive (+) or subtractive (-). This is the universal unit of work across all explorer views.
+ MATCH (c:Concept)-[r*1..1]-(n:Concept) WHERE c.concept_id = 'abc' RETURN c, r, n;
+ MATCH (c:Concept)-[r*1..1]-(n:Concept) WHERE c.concept_id = 'def' RETURN c, r, n;
- MATCH (c:Concept) WHERE c.concept_id = 'xyz' RETURN c;
The + operator merges results into the graph. The - operator removes nodes and their edges. This lets you sculpt a graph by building it up and trimming it down.
Creating a Saved Query
From Smart Search (2D/3D Graph)
Every action you take in the graph explorer is recorded as a step:
- Search for a concept — step recorded as
+ MATCH ... - Right-click a node and choose "Add Adjacent" — step recorded as
+ - Right-click a node and choose "Remove from Graph" — step recorded as
- - Follow a concept (double-click) — step recorded as
+
When you have a graph you want to keep: 1. Open the Saved Queries panel (folder icon in the left rail) 2. Click Save Exploration 3. Give it a name 4. The full step sequence is saved as a replayable program
From the Cypher Editor
Write + and - prefixed statements directly:
+ MATCH (c:Concept)-[r*1..2]-(n:Concept) WHERE c.label CONTAINS 'governance' RETURN c, r, n;
- MATCH (c:Concept) WHERE c.label CONTAINS 'deprecated' RETURN c;
Save via the Saved Queries panel.
From the Block Editor
Block diagrams compile to Cypher statements. Save via the panel as a block diagram definition.
Loading a Saved Query
- Open any explorer view
- Open the Saved Queries panel (folder icon)
- Click a saved query
The system replays each statement in order, applying +/- operators to build the graph state. The result is identical to the original exploration.
Cross-Explorer Flow
Saved queries are shared across all explorer views. Save in one, load in another:
| Explorer | What it shows from the same query |
|---|---|
| 2D Graph | Force-directed node/edge visualization |
| 3D Graph | Spatial perspective of the same graph |
| Cypher Editor | The +/- statements as editable text |
| Vocabulary Analysis | Relationship type breakdown of query results |
| Document Explorer | Source documents for the query's concepts |
| Polarity Explorer | Semantic axis analysis of query concepts |
| Embedding Landscape | Embedding space projection of query concepts |
Switching views preserves the graph. The folder icon appears consistently in all rails.
Query Types
Each saved query has a type that determines how it replays:
| Type | Saved From | Contains |
|---|---|---|
exploration |
Smart search, Cypher editor | { op, cypher }[] statements |
polarity |
Polarity explorer | Pole concept IDs + analysis params |
block_diagram |
Block editor | ReactFlow nodes/edges layout |
The Saved Queries panel shows type-aware subtitles (e.g., "3 steps" for explorations, "2 poles" for polarity).
The +/- Operator Algebra
The operators work like set arithmetic on graph results:
+(union): Execute the Cypher statement and merge its nodes/edges into the current graph. Duplicate nodes are deduplicated by ID.-(difference): Execute the Cypher statement and remove matching nodes from the current graph. Edges connected to removed nodes are also removed.
Example: Building a Focused Subgraph
# Start with everything related to "governance"
+ MATCH (c:Concept)-[r*1..2]-(n) WHERE c.label CONTAINS 'governance' RETURN c, r, n;
# Add concepts about "compliance"
+ MATCH (c:Concept)-[r*1..1]-(n) WHERE c.label CONTAINS 'compliance' RETURN c, r, n;
# Remove noise — concepts about "legacy systems"
- MATCH (c:Concept) WHERE c.label CONTAINS 'legacy' RETURN c;
This produces a graph focused on governance + compliance, with legacy system noise trimmed out.
Exporting to Cypher
From the 2D/3D graph explorer, your exploration can be sent to the Cypher editor:
- Build your graph through search and navigation
- The exploration session is automatically tracked
- Switch to the Cypher editor tab
- Your steps appear as
+/-prefixed statements - Edit, rearrange, or share the text
The text format is designed to be copy/pasteable between users.
Document Explorer Integration
The Document Explorer uses saved exploration queries differently:
- Load an exploration query
- The system finds all documents containing those concepts
- A multi-document concept graph is built automatically
- Passage search adds colored rings to matching nodes
- Double-click a document to view it with highlighted passages
Tips
- Name queries descriptively — "Governance + Compliance minus Legacy" is better than "Query 1"
- Start broad, then subtract — it's easier to remove noise than to find everything piecemeal
- Use the Cypher editor to inspect — seeing the statements helps understand what an exploration actually does
- Cross-explorer replay is instant — saved queries don't re-fetch from the API, they replay the recorded operations
Next Steps
- Exploring Knowledge — General search and navigation
- Querying — CLI, API, and MCP query access
- Polarity Axis Analysis — Semantic dimension analysis