Ontology Explorer

The Diagram Nobody Drew

You inherit a data integration project. Fourteen agencies, each with their own systems, each publishing data described by a different ontology. Your job is to map them together — align the concepts, reconcile the relationships, build a unified model that lets a single query span all fourteen sources. The ontologies arrive as Turtle files. Some are a few hundred lines. One is twelve thousand triples.

You open the twelve-thousand-triple file in a text editor. Classes, properties, restrictions, imports, blank nodes. You scroll. And scroll. You start drawing boxes on a whiteboard, but by the time you’ve mapped the third level of subclass relationships, you’ve run out of space and lost track of where you started. The ontology is technically correct — it passes every validator — but you cannot see it. You cannot hold the shape of it in your head.

Ontology Explorer takes that file and turns it into something you can navigate. Drag and drop the Turtle file, and within seconds you have an interactive graph: classes as nodes, properties as edges, the whole structure laid out spatially so that clusters and hierarchies become visible at a glance. Click a node to see its properties, its superclasses, its relationships. Switch to the hierarchy view to see the inheritance tree. Switch to the chord diagram to see which classes are most connected.

Now you can actually do your job. You can see that Agency A’s Person class and Agency B’s Individual class have the same properties. You can see that the spatial ontology imports three other ontologies that overlap. You can spot the orphaned classes that nobody references, the properties that point to classes that don’t exist in this file, the modelling decisions that will cause problems downstream.

The twelve-thousand-triple ontology is still complex. But now you can see it.

A Map of Your Map

Imagine you’re building a giant Lego model, but instead of a picture on the box, someone gave you a list. A very long list. “Piece 4217 connects to piece 3802. Piece 3802 is a kind of piece 1100. Piece 1100 has a colour property.” Thousands and thousands of lines like that. No picture. Just the list.

You could build the model by reading every line, but you’d go cross-eyed before you finished the first page. What you really want is the picture on the box — the one that shows you what the finished model looks like, where the big pieces go, how the sections fit together.

Ontology Explorer makes that picture. You give it the list (an ontology file that describes how data is structured) and it draws the diagram — boxes for the things, lines for the connections, colours for the types. Now you can see the shape of the data instead of reading about it line by line.

It’s the picture on the box.

From File to Interactive Graph

1. Load your ontology

Drag and drop a file onto the canvas, paste a URL, or paste the content directly. Ontology Explorer accepts Turtle (.ttl), RDF/XML (.rdf), OWL (.owl), N-Triples (.nt), and JSON-LD (.jsonld). The parser auto-detects the format. You can also load multiple files to explore how ontologies relate to each other.

2. Explore the graph

The tool renders an interactive force-directed graph. Classes appear as nodes, properties as edges. Drag nodes to rearrange the layout. Zoom in on clusters of interest. Click any node to open its detail panel: properties, superclasses, subclasses, restrictions, and annotations. Use the filter panel to show or hide classes, object properties, datatype properties, and individuals. The search bar finds any entity by name or URI.

3. Switch views

Three visualisation modes highlight different aspects of the same ontology. Network view shows the full graph with force-directed layout — good for seeing clusters and connections. Hierarchy view shows the class inheritance tree — good for understanding subclass depth and breadth. Chord view shows inter-class relationships as arcs — good for spotting which classes are most interconnected.

4. Detect problems

The problem detector scans for common modelling issues: orphaned classes with no relationships, properties pointing to undefined classes, missing labels, circular imports. Results appear as a list you can click through, jumping directly to the relevant node in the graph.

5. Export your work

Export the visualisation as PNG or SVG for documentation and presentations. Export query results as CSV, JSON, or Markdown for further analysis. The Auto-Tour feature generates a guided walkthrough of the ontology’s structure, useful for onboarding colleagues who need to understand the data model.

Making the Invisible Visible

Data integration scoping

A government department is merging datasets from six agencies into a single catalogue. Each agency describes its data using a different ontology. The integration architect loads all six into Ontology Explorer and uses the network view to identify overlapping concepts. Two agencies define Location differently; three share a common Organisation class but with different property sets. The chord view reveals which ontologies are most interconnected and where the mapping effort should focus. What would have taken weeks of reading specification documents takes an afternoon of visual exploration.

Ontology review and quality assurance

A standards body has published a new version of a domain ontology. Before adopting it, a data team loads both the old and new versions into Ontology Explorer. The hierarchy view immediately shows that three subclass relationships have been restructured and two classes have been deprecated. The problem detector flags four properties in the new version that reference classes not defined in the file — an import dependency the documentation doesn’t mention. The team reports the issue to the standards body before it causes problems in production.

Teaching and onboarding

A university lecturer is teaching a module on knowledge representation. Students submit ontology assignments as Turtle files. Instead of reading each file line by line, the lecturer loads them into Ontology Explorer and uses the Auto-Tour to step through each student’s model. Structural issues — disconnected subgraphs, missing domain declarations, classes with no properties — are immediately visible. Feedback becomes specific and visual: “your Event class is isolated from the rest of the model” rather than “your ontology has structural issues.”

DCAT-AP catalogue inspection

A data portal manager wants to understand the structure of their DCAT-AP catalogue before running a metadata quality audit. They export the catalogue as Turtle and load it into Ontology Explorer. The network view shows every dataset, distribution, and publisher as connected nodes. They can immediately see which datasets lack distributions, which publishers have the most datasets, and how the catalogue’s graph is shaped. It’s a structural X-ray before the quality audit begins.

Under the Bonnet

Parsing

Client-side RDF parsing via a JavaScript RDF library. Supports Turtle, RDF/XML, OWL/XML, N-Triples, and JSON-LD. Files are parsed in the browser — no data is sent to the server. Large ontologies (10,000+ triples) parse in under two seconds on modern hardware.

Visualisation engine

D3.js force-directed graph with custom layout algorithms. Three modes: network (force simulation), hierarchy (Reingold–Tilford tree), chord (radial arc diagram). Layout parameters are user-adjustable: node repulsion, link distance, charge strength, density presets (Compact, Balanced, Spread Out).

Entity filtering

Toggle visibility of four RDF entity types: classes (owl:Class, rdfs:Class), object properties (owl:ObjectProperty), datatype properties (owl:DatatypeProperty), and individuals (typed instances). Search by rdfs:label, local name, or full URI. Filters update the graph in real time.

Problem detection

Static analysis of the loaded graph. Checks for: orphaned classes (no incoming or outgoing properties), dangling property references (domain or range pointing to undefined classes), missing rdfs:label annotations, circular rdfs:subClassOf chains. Results are clickable — selecting a problem centres the graph on the affected node.

Export

Visualisation: PNG (rasterised canvas) and SVG (vector, editable in Illustrator/Inkscape). Data: CSV, JSON, and Markdown for query results and node details. Auto-Tour generates a structured walkthrough as Markdown.

Architecture

Single-page application. All parsing, layout computation, and rendering happen client-side. The Django backend serves the static assets and handles the landing page. No user data is stored. No authentication required. Self-hostable via Docker or any static file server. Open source, licence TBC.