We need to take on board that virtual visitors aren’t just regional, local or real, they’re international and they are just as episodic as someone coming into a museum to shelter from the rain, and they make haphazard but sometimes important use in an instant or in ten years time or in a months time: it’s also international use that’s really interesting.
Frances’ really galvanizing point for me is that museums need to see themselves as belonging to a sector not as individual institutions. We’re not working in a complementary environment and that’s the key to the sustainability issue. There are, however, moves towards getting things interoperable. Something I’m really excited about is Subject Specialist Networks, where people come together by subject not by region. [Jon Pratty, 24 Hour Museum]
Its clear that we need to identify what’s happening inside the institution, as everything isn’t going to fit unless the organizations are the right shape to accept it. [Ross Parry, Dept of Museum Studies, University of Leicester]
We brought together two concepts, one is the whole long tail of how you alternate the internal process. It’s to do with databases and creating digital records and the second model is the delivery of digital services and the use of objects as a way of illustrating narratives on the web. As David (David Dawson, MLA) said, funders will fund projects if you couch them in terms of outcomes and access and learning and that’s blurred the boundaries between those two.
If our small semantic web is what we in our industry do in terms of inside our systems and within our infrastructure and the big semantic web is the world of online service provision then somehow we need to create covalences between our process and the big process and those links that say that our content is going to be disclosable, available and sustainable because of what we fundamentally do with out collections. [Nick Poole, Director, MDA])
The external digitization projects have often been a bad thing as they have reinforced problems and mistakes that we as an institution have made, i.e. unintegrated pockets of automation and we have suffered both internally and externally. We have to watch we don’t make the same mistakes again with grant plans. We have to get museums to push out data from inside rather than getting external money and throwing resources at it. One way to do that is for museums to move away from the concept of applications, packages and buy in systems. Surely that is where the semantic web can be a solution in that we don’t go out and spend 100,000 buying a content management system. (AC, V&A)
What we are talking about is a shift from systems to interfaces. For the pipes to work the data has to flow through the pipe and make sense when it comes out the other end. Then you can put different blocks on the end of the pipe and you don’t mind what block you’re sending in the data as long as your block can make sense of what’s coming through and that the way you get modularity and that’s essential to the idea of web 2.0, its machine processable information arriving somewhere and being useful, and that brings us back to interchange formats. (Richard Light, consultant)
You (Alan) implied that collections management was already collection shaped. It’s not that you can see how we can make things semantic, it’s that you want them to be that shaped? (RP)
It’s what Suzanne and Richard were saying. You send someone some data and they do what they want with it. If they want to use it in a narrative, they can do so, if they want to use if for some other purpose, then that’s up to them. We’re still in the mode of expecting people to use data in a certain way. (AC)
If we look at how others publish successfully and how they join up successfully in other spheres, it’s always very low technology, it always doesn’t involve complicated, consensual, self-referential mechanisms. On one level you might say perhaps things like mash-ups etc, that that seems to be a model for what we’re talking about. Museums aren’t seeing that they need a wider publishing strategy as part of their practice. (JP)
We need a strategy as institutions where we see ourselves in this information space. (RP)
Estate agents are a good example. There are about three major sites nationally that skim the data from other estate agents. The same with train timetables, they skim their data and present it in a better way. (Suzanne Keene, UCL)
The estate agents is a good example, it’s about branding. That local branding has no relevance to someone searching internationally, and that’s something that institutions haven’t got their head around. (David Dawson, MLA)
We’re not apparent in all of the aggregated hotel booking services, we’re not apparent in the tourist information bureau websites, we’re not apparent through the English tourist board website and actually in terms of aggregate investment that all that would make you get a far greater audience than you ever did creating a door marked museum. The value of all of this has to be through third-party disclosure cause that’s how you aggregate a larger audience. (NP)
Is this project consistent with content management? (RP)
I have a system diagram of the major chunks of content management. When I looked at it I couldn’t see how semantic web technology couldn’t help any of them. I think it could greatly assist the practicalities of content management. (SK)
You start with your legacy information, your collection and you enrich this information to a semantic level and interlink them on legacy level. Take the domain experts and let them interlink the information so you end up with better interlinked information.
No longer just describing objects and the only information that links them is that they were in the same exhibitions some time ago, or they are by the same painter. There is some more information that Dublin Core delivers. However, the problem is that delivering this legacy level is very expensive and it will take some time to mash it all together.
Usually, what we’re struggling within digital libraries at the moment, is what you end up with is partially linked up information cause you don’t have enough people to sit down and start interlinking and you don’t have smart engines to do it for you. The social networking side is the future because the users who visit the museum remember that what they are seeing is relating to something over there and they start tagging, creating their own collections.
People can classify how they see the work. The soft semantic web and the hard Semantic Web are just different layers – the foundations, the legacy semantic web and the users build upon that. (SK)
Whereas, in the 1970’s we came up with a data model that was initially hierarchical and then became potentially relational as well, to create our records management system which facilitated access, what we’re doing now, thirty years later, is we’re coming up with another data model which is more object orientated, which is facilitating discoverability. It’s not a records management system, it’s all about facilitating access.
I think semantic web, whether it’s upper or lower, it does three things:
- it facilitates more efficient and intelligent discoverability. It makes the web work better for you. You can find things when you search for ‘watch’ that are about a timepiece and not about the verb. It may also suggest things that the web suspects are probably relevant.
- It thinks in terms of resources not pages. It grinds things down into cultural atoms, the granularity of stuff. People aren’t discovering a page, but a news story, an image, a label.
- Its about machines, computers talking to each other. It’s about me searching for a book, an object in a museum collection, and the computer coming back to me and saying that object’s in the V&A, the V&A is open on Tuesdays10 until 5, you’re free 10 to 5 tomorrow and you could get this train. It could link across a number of datasets.
There’s a hard version that may require a framework that sits above the collections, with an agreed ontology to make all that work and there is a softer version that happens in a more serendipitous, from below, fluid way. (RP)
I think that there is a key question for us in this, as far as I understand there is a difference in the semantics of recording collections and the semantics of using the collection and this is the key problem the semantic web faces.
That when the semantics of recording are tweaked but they are not perceived as the same by the user. The semantics of use are completely free, there is no structure we can impose of the semantics of use. We are trying to put together two aspects that are in conflict.
There are semantics of recording, and they are shared across institutions, for example the subject specialist networks try and create the semantics but it doesn’t mean they will overlap with the semantics of use. When we talk about loans, or space, or finances, we are going back to what we said in Glasgow – that some aspects of the function are more appropriate or easier to be semantically coded, as when in Brighton we talked about microformats.
We know our address and our postcodes and the dates of specific exhibitions and semantics that are more-or-less universally accepted, therefore, everyone was really quite excited, but when it comes to collections we have very different ways of recording and using them. How can we make these stick together? (Areti Galani, University of Newcastle)
It’s a really lovely distinction between the semantics of record and the semantics of use. In a way we’ve got tied up a little bit in the last generation because we’re trying to use the UK Spectrum, the semantics of record – On the one hand there’s the semantics of record, that’s internal, that’s part of the joined up culture and then there’s the semantics of use which is about discoverability, user generated content. (RP)
What’s interesting here is that we’re seeing a mixture of chaos and order. The one thing that is happening at Powerhouse is that every single object on the database findable by the search engine – exposed the whole thing and tagged everything – and it’s transformed the usage of the database.
And that’s a good thing because it means that the findability and useable, this long tail, millions of fragmented users using the collection in millions of fragmented ways works brilliantly. But, as a curated experience, a themed experience, or as a series of clumps of stuff joined together in narrative form, it’s really hard to use.
So we do see, in Powerhouse particularly, that we now we need to give users a slightly different way, a curated way of experiencing clustered, tagged, conglomerated stuff. (JP)
There was some research done recently on this tag and folksonomies. Everyone started to agree that users are different and come up with different words– when we look at tags they really mean something different to different users and unless we are able to extract the meaning – create another tag underneath them – then we are unable to use them to the extent we would want to. (SK)
I’m really not in favour of tag clouds…they are really flat. We need to deliver something nicer for users. Tagging information by users shouldn’t be restricted to the objects themselves, they should be able to zoom into smaller parts, areas of interest. (SK)
But it treats people using delicious as a community, so for example, if you put in a tag for the powerhouse museum you would offered tags that other people have used to identify it and if you clicked on it you could go and see the other things that they are interested in. A useful way of using this would be to invite some sort of online community to address a particular area of the collections and that community could jointly tag up the collection with what they found interesting. (SK)
When people are being semantic, they shouldn’t know that they are doing it. (NP)
1) funding is not consistently core for this work in museums
2) Semantic Web logic is consistent with collections management
3) (sorry – it cut off – don’t know why) (RP)
Sebastian Kruk from Finland did a demonstration of the JeromeDL SW demonstrator. [more details of this will be uploaded later – JP]
I can see how this is drawing capabilities, disparate, separate things together in different forms to suit different users. (AC)
Exactly, in Jerome, different parts come from different projects some is taken from in-house, small building blocks that you can use in different ways. Through Jerome we can pull it all together and build a stable platform, but if you don’t like it, use these components and build your own stuff. (SK)
Collections Management Systems (SK)
Its not just about registration, its not just about archives, and its not just about cataloguing collections, it (the diagram) shows a lot of the processes that go into a collections management system, but I’m sure we’ll also think of other ones. I thought It might be useful for us to think about how semantic web technologies can help each of these processes.