Raising the stakes on ontologies
Posted on 6 July 2012
Raising the stakes on ontologies
By bringing together experts in ontology development and digital preservation across a number of fields, the Software Ontology Project (SWOP) gathers useful descriptions about software in an ontology called SWO1. The laudable goal is to tailor the SWO to make a useful ontology that is also accessible to people without expertise in ontologies. Having been to the first two SWOP workshops, I was delighted to be invited to the third workshop last week.
Working mainly in pairs, we added descriptions of software to a uniquely customised Excel spreadsheet that represented the SWO. A neat technology called RightField, which is developed and maintained by the SysMO-DB team, allowed us to rapidly add a plethora of software annotations in Excel. This means that anyone can use the SWO ontology, even if they don't understand its complexities.
The next step was to look at 'competency questions' of the SWO itself. Everyone came up with questions based on what they'd use the SWO for: 'What input data formats are supported by X?', and 'What equivalent software exists that could successfully replace X?'. These questions were separated into categories (data formats, support, cost, features and suchlike) and, interestingly, the questions were far easier to delineate than the last time we performed this exercise. In effect, this means we've developed a much simpler classification!
The next part was a great deal of fun: Priority Poker. It's a useful way to decide on what categories to capture within the ontology, given the limited amount of effort and resources that are available. The first step is to score the categories: each person gives each category a score from 0 to 100 dependent on how difficult they think it will be to capture the category in the ontology. The scores are totalled to give each category an overall measure of "difficulty to capture". The next step is the buy out: each participant is given points that can be used to buy categories. If you're lucky, you can buy the categories that are important to you using the points you have. But more often than not, expensive (i.e. difficult) categories require you to persuade and cajole others to pool their points with you. In the end, we had list of categories that had been bought and would be included in the SWO, and a list of runners-up that had automatically been prioritised.
Priority Poker is a very clever way of promoting discussion and generating a collective decision for what is important. It's an activity that I'll definitely be using again.
Following the workshop, the information that was gathered at the workshop will be used by the SWO team to rebuild the SWO using Populous (which is based on RightField). This approach allows a virtuous cycle of ontological data gathering and improvement of the SWO.
My thanks to Robert Stevens, James Malone and Helen Parkinson for the invitation to the third, and unfortunately final, SWO workshop.
[1] Maria Copeland, Andy Brown, Helen Parkinson, Robert Stevens and James Malone. The SWO Project: A Case Study for Applying Agile Ontology Engineering Methods for Community Driven Ontologies. To be published in Proceedings of the 3rd International Conference on Biomedical Ontology (ICBO 2012), Graz, Austria, July 21-25, 2012.