Visual Graph Explorer

Overview

The Visual Graph Explorer is a read-only Dataiku webapp designed for business users and analysts to perform in-depth visual exploration of graphs exposed through the webapp.

These graphs can come from Visual Graph-published folders and, when configured, unmanaged Neo4j databases.

Note

Get started by following this Explorer webapp tutorial.

Key features

• Schema overview

  View the graph’s structure, including all node and edge groups.
• Visual exploration

  Visualize graphs with different layouts, expand neighbors to explore relationships, and apply conditional styling based on node properties.
• Saved queries

  Execute a library of saved Cypher queries prepared by data scientists to answer common business questions.
• Ad hoc querying

  Write and run custom Cypher queries directly within the interface with schema-aware autocompletion.
• Query generator

  Use a built-in LLM assistant to generate Cypher queries from natural language questions.

Note

All schema design, data mapping and visual customization are performed beforehand in the Visual Graph Editor webapp.

For unmanaged Neo4j databases exposed directly in the Explorer, the webapp infers the available schema from the database and does not rely on a published Visual Graph configuration or on a graph being materialized in a Dataiku Folder first.

Settings

Graph databases

• Folder(s) containing the graph databases: select one or multiple Dataiku Folders containing graph databases published from Visual Graph. Leave this empty if you only want to expose unmanaged Neo4j databases directly in the Explorer.

• Neo4j connection: if some selected folders represent a Neo4j graph database, or if you want to expose unmanaged Neo4j databases built outside Visual Graph, configure the connection to be used by the webapp.

• Unmanaged Neo4j databases: select one or multiple unmanaged Neo4j databases to expose in the Explorer through the selected Neo4j connection. No Dataiku Folder materialization is required for these databases.

Note

Only one Neo4j connection can be used for all selected folders and unmanaged Neo4j databases in a given Explorer webapp.

To access Neo4j databases available through different connections, you will need to create separate Explorer webapps for each connection.

AI assistant

• LLM connection to use (Optional): select a LLM connection to enable the AI-powered Query generator. It helps users construct complex Cypher queries using natural language.

• LLM history dataset (Optional): select the dataset storing the history of all questions asked by users in the Query generator. It can be used for auditing or analysis.

Advanced settings

• Cypher query timeout (seconds): set a maximum execution time for queries run within the Explorer. This acts as a guardrail to prevent resource-intensive queries from impacting performance.

• Log level: configure the verbosity of the logs. Select INFO for standard operational logging (recommended for production) or DEBUG for detailed diagnostic information while troubleshooting issues.

Note

Not configuring optional settings will simply disable the associated feature.

Visual Graph Explorer Interface

Landing Page

The landing page displays a list of all available graphs exposed through the selected graph database folders and, when configured, the selected unmanaged Neo4j databases.

For graphs published from the Editor, metadata such as the last build time is shown to indicate data freshness.

Graph Exploration Page

Left-side panel

Graph information

The top of the panel displays the graph’s name.

For graphs published from the Editor, it also displays the last build time and any comment saved with the published configuration.

Groups

All node and edge groups are listed with their respective colors.

Graph statistics

For each group, two types of counts are shown:

• visible: The number of nodes/edges currently displayed.

• total: The total number of nodes/edges for that group in the entire graph database.

View controls

For each node and edge group, you can toggle its visibility on or off to show or hide all elements of that group in the current view.

Bottom panel

Saved queries tab

The Saved queries tab contains any pre-built queries saved with the published graph configuration. They can be executed to see the results rendered as a graph or a table.

Note

These queries are prepared by Data Scientists in the Editor webapp.

For unmanaged Neo4j databases exposed directly in the Explorer, this tab is typically empty because no Visual Graph saved queries are attached to that database.

New query tab

The New query tab provides a powerful editor with schema-aware autocompletion to write and execute your own Cypher queries.

Query generator tab

For complex queries, the Query generator tab allows you to ask a question in natural language. The AI assistant will translate your question into a Cypher query that you can then execute.

Central panel

In the Graph view, you can click on any node or edge to inspect its details, including its identifier, name, and properties.

When a node is selected, an additional Neighbors tab is available. It provides an overview of all directly connected nodes, displaying the total neighbor count and a breakdown by node group.

From here, you can selectively expand all neighbors or only those belonging to a specific group to explore the graph’s structure interactively.

Graph layouts

Switch between different layouts to optimize the visualization of your graph based on its structure.

Options include gravity (the default layout), tree for hierarchical data and circle or square grid grouping to cluster nodes per group.

Warning

Tree layout is computationally intensive. It is not recommended for large graphs.

Conditional styling

Enhance the visual analysis of your graph by styling nodes based on the values of a selected property.

You can define color, size, or icon rules per node group. For numerical values, you can work with ranges and gradients. For categorical values, you can assign styles to specific values.

It helps identify patterns, clusters, or outliers within the graph based on specific attributes.

Note

Conditional styling is a view-level aid for exploration. It does not change the underlying graph configuration, and it is only available for node groups that currently have displayed nodes and available properties.