← Back to list

Graph-Based Visualization of Domain Facts

A picture is often said to be worth a thousand words. In many domains, knowledge is best conveyed as an image, or a directed graph, to be…

Petr Hejda · 2025-11-21 15:18 · 0 claps · 2.1 min read
#knowledge #data-visualization #structured-data #data-engineering #data-science
Open on Medium ↗
Wiki topics: ML · Machine Learning VIS · Visual & Graphic Design 🔧 · Data Engineering 🔬 · Science · General

Graph-Based Visualization of Domain Facts

A picture is often said to be worth a thousand words. In many domains, knowledge is best conveyed as an image, or a directed graph, to be more precise. Visual properties such as layout, color, and node shape help readers understand the structure of a domain, the concepts it contains, and how those concepts relate to one another.

To illustrate this, consider a simple example. The input is a text file describing facts from graph theory. Each line encodes a fact in the form “subject predicate object”, with optional named attributes for any part of the statement. A full sample file is available here.

Text with the facts

Text with the facts

We can use the Jumble Tool to visualize these facts as a graph. Once the file is loaded, the tool displays the data on the right side of the screen. Subjects and objects become graph nodes, and predicates appear as edges connecting them. At this stage the visualization is still minimal, but we will refine it shortly.

Facts loaded to the Jumble Tool

Facts loaded to the Jumble Tool

On the left, the digest summarizes the structure of the data. All nodes sharing the same label are grouped into a single concept, and all edges of the same type are grouped into a single relationship.

The Jumble Tool provides context menus to fine-tune the graph’s visual presentation. A clear and accurate layout is essential for understanding the knowledge encoded in the data.

Context menu

Context menu

There are several aspects of the graph we can refine:

  • Set the top-level layout to free.
  • Change the default node shape to text.
  • Remove the l property (line number) from all relationships.
  • Select all relationships and adjust arrow and text colors.
  • Set the line color of all is relationships to white.
  • Remove the l property from the entity concept.
  • Remove the entity label.
  • Apply distinct colors to key concepts (set vertex to blue, structure to red, link to orange, and attribute to green)
  • Manually adjust node positions for clarity.

The resulting graph reveals a clean, well-structured view of the domain, with clear relationships between the core concepts.

The clean view of the domain

The clean view of the domain

Explore More

To learn more or experiment with the data yourself, check out:


메타데이터
post_id
ee0bcd3bfd04
slug
graph-based-visualization-of-domain-facts-ee0bcd3bfd04
url
https://medium.com/@heyda/graph-based-visualization-of-domain-facts-ee0bcd3bfd04
canonical_url
https://medium.com/@heyda/graph-based-visualization-of-domain-facts-ee0bcd3bfd04
author_url
https://medium.com/@heyda
status
ok
fetched_at
2026-07-15 08:53:23