Tawny with Protege

While Tawny allows construction of any OWL ontology, it has a programmatic interface. By itself, it does not provide strong support for visualisation of ontologies; this is something that Protege is very good at. Here we describe three ways to integrate with Protege, using it as a visualisation framework.

Caveat

First, we start with an entirely negative point. All the uses of Protege that we describe here are good only for displaying the generated OWL produced by Tawny. As tawny can be to build arbitrary abstractions on top of OWL, and is fully programmatic, this may or may not be a useful thing to do. Trivially, for example:

(doseq [k (range 100000)]
    (owl-class (str k)))

will produce a very large ontology from a very small statement. This also means that Protege is read-only -- changes made in Protege will not affect the Tawny source. Our experience is that Protege is actually fairly useful as a read-only viewer.

Caveat 2

At the time of writing, the current release version of Protege uses a different version of the OWL API. So inconsistencies between the two are possible. For our use, we have developed against a development version of Protege patched to use the same version of the OWL API.

Tawny is likely to remain closely synced to releases of the OWL API, rather than Protege, so this may always remain a problem. From a development point of view, building Protege from source is relatively straight-forward, but this might be an excessive demand for "normal" users (and if you know of any, please tell me!).

File-Based Access

The easiest way to use Protege with Tawny is simply save the ontology to a file. This example from tawny.pizza shows how to add this to a main function, which will run with lein run.

(ns pizza.core
  (:use [tawny.owl]
        [pizza.pizza]
        [pizza.render-pizza])
  (:gen-class))

(defn -main [& args]
  (with-ontology pizzaontology
    (save-ontology "pizza.rdf" :rdf)
    (save-ontology "pizza.owl" :owl)))

Now, pizza.owl can be opened in Protege. It will detect when the files have updated, and offer a reload. This workflow is okay, but does not provide a rich interaction between tawny and Protege; in particular, effectively resets the GUI, leaving which is painful.

Tawny in Protege

And second approach is to use Tawny inside Protege. Protege-Tawny provides a console and REPL inside Protege. It is possible to type commands that have immediate affect, on the Protege GUI.

While this works, it essentially offers the opportunity to use Tawny as an OPPL like language for adding new terms to an existing ontology. The source is not saved, and not available next time around.

Protege as an nREPL Server

It is also possible to use Protege as nREPL server. Essentially, this means instead of launching a REPL using your IDE, you launch Protege and then connect to it with your IDE.

This provides the tightest integration with Protege; Tawny code can be evaluated as normal, and any changes appear immediately in Protege. Preliminary integration with Emacs is also provided, enabling "Jump to" and "Tracking" functionality. Currently this an early release (as it requires building Protege from source), but our hope is this becomes stable and easy to use in the future.