Meeting on ontology modelling
Posted in Technology on 2006-09-26 20:50
Today there was another meeting of the Topic Maps Users' Group on ontology modelling in Topic Maps.
Ontology Modelling Methodology
I started out with an early version of my TMRA'06 presentation, which is about a methodology I've created for Topic Maps ontology modelling. The methodology is basically in two parts: a process for creating ontologies, and a set of guidelines for correct modelling. The purpose of the process is to help modellers take all the various requirements and project constraints into account and produce a model that is well understood by all project participants.
The guidelines are basically a set of rules for how to use the Topic Maps model correctly when modelling the domain. The guidelines are effectively the rules taught in Ontopia's ontology modelling course (what's known as day 2). These are unfortunately not written up anywhere yet (except in the slides), but the paper will have a good subset of these. Some have of course also been debated in this blog already, and more will probably follow as time allows.
I was followed by Eszter Horvati (of Ovitas) and Graham Moore, presenting HistoryMaps, which they referred to as a spare-time project. The idea is to provide an ontology that allows people to use Topic Maps to model historical events; the inspiration is Kal Ahmed's PepysMap project, which modelled Samuel Pepys's diary as a topic map.
The basic model is event-centric, and provides various axes along which events can be described. The most obvious are start-time and end-time; more contentious are "organism" (organization involved), "subject" (people involved), "artifact" (manufactured items involved), "theme" (categorization using Dewey), "location" (geographic). In addition, descriptions and references can be attached.
Two examples of events were provided: "Columbus arrives in San Salvador", and "Colombus establishes settlements in Santo Domingo". I asked if "the Industrial Revolution" would count as an event in this ontology, and the answer to this was "yes".
The presentation of the model sparked vigourous discussion about what the granularity of events is, what the difference between a state and an event is, how to deal with fuzziness in time, how to account for people having different interpretations of events, etc etc. Graham and Eszter made it clear that different interpretations were supported simply by having different event topics, and that the model in general is meant to be a "loose binding" that does not enforce people to buy into a particular historical view.
Graham summarized the whole thing quite well by saying that HistoryMaps had proved to be "a good exercise" in the sense that it really got people engaged and sparked discussion; in short it was "good fun".
Later discussion with Graham and Eszter over beer revealed that they actually have registered the domain name, and are collaborating with a historian (Sturla). They've built an editing application and plan to go live with this as a collaborative project a la Wikipedia. It will be interesting to see how this goes.
Are Gulbrandsen was the last speaker with a presentation on how they (USIT) had built the ontology for the Houdini project. He showed how the ORM modelling formalism was used to build the ontology. (His TMRA'05 paper was about modelling Topic Maps with ORM in general.)
I didn't catch too many of the details of this presentation, unfortunately, as I was too busy typing up the notes from the two previous presentations (Eszter borrowed my laptop for her presentation).
After this, we adjourned for beer. The next meeting will be October 24th, and with luck I'll blog that, too.
This is a semi-live report from the first Topic Maps conference: TMRA'05 in Leipzig
Read | 2005-10-06 16:33
Read | 2008-10-26 17:30
Lars Marius - 2006-10-13 15:52:57
In case anyone gets confused by this: the user group referred to is the Norwegian Topic Maps Users' Group.
Robert Cerny - 2007-06-11 12:53:36
Is the HistoryMaps ontology topic map available online?