Presenting: the ODU-browser


Experimental; mostly raw ontology data for now. Protégé will read this from the URL http://www.den-uijl.nl/ontologies/den-uijl.owl.

There's an early, incomplete, experimental, designed at the keyboard, and all that, graphic ontology view- and search interface in the making, that I named the Ontology Drawing User-interface, or ODU-browser for short ;-).

The idea is to create an ontology viewer / search interface that runs as a fast thin client in a browser, can be operated on pads with one finger, doesn't need too much screen real estate to keep a good overview (which is where a lot of ontology viewers fail), that will interface with OWL-files and triple stores, and that can be used as a personalised search interface by simply setting up a screen you like and saving that view.

The content that is searched doesn't have to be on the same server as the ontology, as long as the ontology has information on it. This means that if several parties use the same ontology, or taxonomy and thesaurus, for adding metadata to their content, and if they put their "part" of the combined but distributed ontology on their servers behind a standardised API, they could do integrated search and retrieve documents without the burden of having to index each other's content or exchange indexes or other data.

I would be interesting to include content and its relationships into an ontology, creating a browsable "cloud of organised knowledge", and integrating that into the viewer, blurring the customary division between ontology and content, and storing that in a distributed way, so that any individual user can be his own content integrator, without the need for a third party integrator. It would require a shared view and implementation of the associated ontology.

The ODU-browser is based on the combination of SVG and JavaScript, mainly because I hadn't tried that before. It reads only what it needs in small bits from a server, putting the query load there. At the moment, the underlying ontology is an OWL XML file that is queried by a PHP page. I have also implemented an interface to a relational database, which greatly speeds up things. Querying more than one, or a distributed, ontology is only more of the same.

So far, I have implemented classes, superclasses, subclasses and instances, and a search mechanism that gives search results from the ontology and from Google's site search. I have also implemented thesaurus-like relations between keyword instances like broader terms, narrower terms, related terms, and synonyms.

Note that this has only been tested in Firefox. Opera, Chrome and Safari seem to run it as well. IE8 doesn't run it, I don't know about IE9 (and I don't really care). Also note that I haven't implemented touch screen gestures yet, so use a mouse or touchpad for now.

This page was created by Oscar den Uijl, oscar@den-uijl.nl