Mia wrote:
I'd like to be able to link to other organisation's people and place
> authorities/information records so I could set up Wikipedia-style inline
> or sidebar-style 'related' links to help visitors browse across museum
> sites. Not every person we think is interesting is 'notable' enough for
> a Wikipedia entry, but across the UK we must have lots of records about
> the same people, events, places, etc.
Interesting. We've been building a tool called Muddy (http://www.muddy.it/)
which identifies entities in content using data from DBpedia, and does some
clever disambiguation tricks. It's currently being used internally by the
BBC, as they have decided to settle on using mainly DBpedia as an authority
source/'equivalency engine' (see diagram http://twitpic.com/il1w).
DBpedia is pretty powerful, and getting ever more so, but does rely on
Wikipedia for all its data. This has some benefits in that Wikipedia can act
as a filter for 'significant' references (people/companies/places/etc),
however it does also mean that theres a notability threshold which may not
correspond to your content.
I see two ways round this.
The first is to find some other domain-specific open content source, like
MusicBrainz (http://musicbrainz.org/ - a great resource for bands and
albums) and possibly OpenLibrary (http://openlibrary.org/). See Michael
Smethurst's blog post (and the comments) for more about this
http://www.bbc.co.uk/blogs/radiolabs/2009/01/in_search_of_cultural_identifi.shtml
The second is to actively contribute back the data you have, either by
providing your own open data library (like the BBC's programmes site), which
is fairly complex and high-risk, or by adding it to Wikipedia (if you think
it DOES pass their notability criterion, which should be the case for any
artists exhibited in a major gallery, for example) or alternatively
Freebase, which doesn't have any notability requirements. I've started to do
this around exhibitions (see http://exhibitions.freebase.com/).
The important thing is that whatever resources/references you use, they
should be open, free, and 'permanent' - which normally means being managed
by open communities rather than institutions. This means you end up, in
effect, 'using the web as a cms' (ref: Tom Scott,
http://derivadow.com/2009/01/13/the-web-as-a-cms/).
This might all sound geeky and technical, but once this groundwork has been
laid, we can start moving beyond cross-institutional keyword searches and
start enabling properly linked data sets that are part of the 'web of data'.
Cheers,
Frankie
P.S If anyone wants to find out more about our Muddy tool, or to have a
play with it, let me know.
--
Frankie Roberto
Experience Designer, Rattle
0114 2706977
http://www.rattlecentral.com
Sent from: Sheffield Sheffield United Kingdom.
**************************************************
For mcg information and to manage your subscription to the list, visit the website at http://www.museumscomputergroup.org.uk
**************************************************
|