This final point from Seamus led on to a discussion about demonstrator projects and his recommendation that the Semantic Web Thinktank should not focus solely on producing documents and publishing reports and rather disseminate the outcomes through projects and demonstrators. Others, however, expressed their concern with this approach, Mike Lowndes, for example, explaining that the outcomes of demonstrator projects often limit the vision of those that are being demonstrated to.
Similarly Jon Pratty was wary of demonstrators, noting that many of the successful projects that exist on the web do not have the word demonstrator wrapped around them. To him successful projects were more organic and were produced ‘relatively cheaply, relatively quickly and instinctively’.
This generated a wide-ranging discussion clarifying the intentions of the Thinktank. While many concurred that there was a serious need for a tangible outcome there was some debate on whether this should be a demonstrator or not. Some suggested quite simply that the word generator should not be used. Others suggested alternatives, such as ‘sand box’ a term for a more fluid, open and interactive project that people could play around with.
To others, however, the tangible outcome would come more in the form of recommendations or an agreement on standards, whether this would be agreed on during the Thinktank or whether it would build incrementally over time. This would have the advantage of not being very expensive and having long-term impact as Nick Poole explained.
‘What’s needed is a decent standard – you want Spectrum or something like it to be built around semantic ontology structures which then have a long term impact. If you do that then collection management systems start to be that shape and people’s training starts to be that shape…’
Richard Light suggested:
‘I can imagine a project which we could build with actual content from a set of museum databases which comes in in an agreed Semantic Web format into some central place and uses standard Semantic Web standards for representing it and you could do something like the meta search project which is available to anyone who wants to play. This way you end up with a central resource.’
Two existing projects were then brought to the attention of the group, The Museum Network and The National Museums Online Learning Project, for the purpose of illustrating collaborations between institutions with an interest in sharing practice and audience and potentially records or information. Jon Pratty suggested that these collaborations, rather than being about generating content, could be about the technologies that allowed them to share and the taxonomies that they would need to develop in order to make this happen.
What became clear after this discussion was the multiple and subtle levels on which the outcome of the Thinktank would have to work. Jeremy Ottevanger’s concern was that the outcome be integratable into what the sector is already doing bringing us back to standards and documentation.
Nick Poole: “One thing that constantly amazes me about our sector is how often it comes down to a single page of A4 paper and the moment of knowledge capture and creation largely comes down to that accession form which is one side of A4 and hasn’t changed in five years. And its very basic information and yet for most museums that’s as sophisticated as knowledge capture gets. So, at the very basic level, if you want to affect long term change we have to change that form. There has to be a moment of knowledge capture that supports a semantic approach.”
The potential of the Semantic Web to increase the interoperability of documentation information was then explored further and paralleled to Richard Light’s work in the 1970’s.
Richard Light: “I think that the data structures that we invented then are certainly fit for purpose for anything that I can see us doing now in the Semantic Web context. And I think the key aspect of that is about describing relationships between things and I think that’s an important thing to bear in mind because, for example, in the current Spectrum Units of Information you basically have a big bag and you open the bag and out come all these marbles and they are all independent units of information – and you go to any metadata source and you get a few more marbles, individual items of information. You don’t get context and you don’t get what I call co-contextuality.”