As some of you might remember, I put together www.museumcollections.org.uk a while back to demonstrate what could be done for collections searching with next to no cash (a fiver to register a domain), time (20 mins, tops) or effort (cut and paste). Underneath this is Google coop, an implementation of the big G’s search engine which lets you search across multiple websites. In this particular example, I added a bunch of domains or sub-domains featuring museum collections and also asked people (so far about 20) to contribute if they wanted to add further domains to the list.
O’Reilly radar posted last week about Open Education Search, a collaboration to “build a web search portal dedicated to open educational resources“. There is more about the project on this later post, but it looks as if it will make extensive use of Custom Search, another offering in the bewildering array of free search services provided by Google.
For me, the interesting thing about using Google coop is that it places the bar for cross-domain collections searching, and automatically challenges any institutions considering the various approaches favoured in the past (such as, for instance, Z39.50), to come up with something better.
There have been rumblings in the pipeline for as long as I can remember about a national (or international) search engine for museum collections. Pretty much everyone agrees we’re in a ridiculous place right now: you have to know which institution to go to in the first place to then do the searching for the thing you’re interested in. There is no central place for finding all the Babbage-related collections on the web, for example, except for – oh, hang on – Google.
“…wait!…” shout the hardcore metadata types, “…Google apps doesn’t provide our users with the granularity they require: we need it to be better!”
Well, here’s a proposed solution to that problem: instead of a bunch of museums getting together and spending the next five years (and equivalent vast sums of money) arguing about standards, interoperability etc, before eventually self-imploding and deciding it’s all too much like hard work, how about we club together and buy a Google Enterprise or two (~£15k education price, I believe) and point it at each of the collections websites. Tweak the results, pay a designer £5k for the end result, buy a domain?
I’m being slightly fatuous (imagine!) but there’s a serious point here: Google does search really, really well, so why not use it? Yes, it’s “brute force” searching, but nothing – nothing – has come even close yet to doing it better. This is a perfection gene issue: I vote for cheap, cheerful and 90% perfect (and actually getting it done) rather than 99% perfect and still being here, £3m worse off and with nothing else to show in a few years time.
So. Anyone got £15k?
I’m not convinced by the granularity argument. We have an advanced search form for our online collections database – it gets less than something like 50 users a week according to Google Analytics.
We’ve just bought the business version of Google custome search, and written a little module that integrates it into our CMS. So you can now visit the Maritime Museum site, search for everything about the ‘Second Dutch War‘ (say) and get back results from our online catalogues, journal articles and general stuff in the CMS. Much better than the old search, where you had to visit the search boxes of the various different sites in turn.
Hi Mike
Yeah I agree . . . we use a Google Coop for our site search (but NOT our collection search – although it also can work as a way into our objects of course!). See it at http://www.powerhousemuseum.com/search/ (it is also in our persistent footer)
And we should have a statewide cross-institutional prototype of a combination Google Enterprise + OAI + Opensearch model working in the next 6 months. I might have a beta up before Xmas.
The problem that I’ve found with Google Coop is the inability to do the kind of deep data mining that we do on our (self built) OPAC search. Effectively that data value goes to Google (and doesn’t come back to us).
There is the other problem that I noticed when I was over in the UK, that SEO seems to still be a ‘mystical thing’ that hasn’t been applied properly to quite a lot of museums’ collections . . .
Seb
Yeh, have to agree with Seb here, the vast majority of ‘detailed’ online Museum collections just aint visible to Google. Doing that itself is a vast task, which is why I and others were pretty excited about dublin core as a basic minimum interoperability standard, oh, years ago.
It gave people building systems something simple but structured to aim at – instead of the ‘voodoo’ that Google SEO can be.
It has its problems. When I developed the Data Locator idea at the NHM, we realised that DC didn’t fit that well and ended up using something even simpler (and user oriented) to map all our online collections data into.
Theres also the problem that different museums name things differently and order collections differently. Mapping them was/is tricky, leading to incomplete cross-collection query results, Thats what led me into the Semantic stuff – tools to aid those mappings (Cue arguments on social tagging!).
So, sorry Mike I have to disagree on this one. Museums etc as a sector mucked up bigtime by not building a national ‘unit level’ metadata repository (that itself could perform SEO!). I know MLA are trying, but really not hard enough!