Finally I’ve started to see more museums picking up the absurdly easy to implement Opensearch method of delivering a live search result from their website as RSS/XML.
The National Maritime Museum in the UK is one who has recently made their implementation of Opensearch available. Here’s a feed of a search of their collection for compasses.
Here’s a feed of that same search for compasses but from the Powerhouse Museum site.
Anyone who wanted to make a website about compasses could easily do so by simply combining these and other search feeds – a search mashup if you will. This is an extremely easy lightweight way to implement a cross-collection search – and whilst it does have significant scalability (how many institutions’ compasses do you want to combine?) and ranked combining issues (whose compasses do you show first?) – the actual technical implementation, if you run your own collection search already on your website are extremely trivial.
We’ve had Opensearch here at the Powerhouse for two years now and it provides slow and steady trickle of users. Libraries Australia uses our feed to provide a Powerhouse search within the libraries portal all over the country, and the Collections Australia Network and the Museum’s own dHub use it to offer a Powerhouse search on their sites.
So, why haven’t you implemented it yet? Or, if you have, let’s start sharing our feeds.
Our Opensearch description file which specifies the syntax for queries to return results either as HTML or RSS/XML is located here. Feel free to use it and tell us about your creative mashups.
11 replies on “Opensearch – it isn’t all that hard”
Thanks for the link, Seb.
I’m curious to know why haven’t you made your OpenSearch feeds easily discoverable from your website?
And do you have any structured feeds that use, for example, your categories, tags or maker, rather than search queries?
Hi Fiona
Having done the initial implementation two years ago we generally felt that Opensearch is for machines rather than humans. Collections don’t change that much so the idea of ‘subscribing’ to a search result seemed a little ridiculous especially compared to subscribing to say a search on a news or sports site.
That said, we’ve always advertised the Opensearch to browsers in the header of our collection pages (link rel=) and we will add a RSS link logo in the next few days – mainly because now more museums are doing Opensearch and consumer level RSS acceptance/use is on the rise.
The only RSS feed of collections that we intend to give any real prominence to is ‘new acquisitions’, which we’ll be launching soon.
However, we do photograph and publish new records for at least 1,000 objects a year at the NMM and we suspect that academic and family-history researchers might be interested in subscribing to receive updates, particularly by type/ person/ ship/ site…
We also expect that providing an email alert option (e.g. via Feedburner) would significantly increase usage.
As an experiment, I had a go at running one of our new feeds through an RSS-to-KML converter and displaying it in Google Earth. I posted a screenshot on flickr. Pretty cool, I think.
One potential use of feeds that I thought of – aggregating together all the material from Cook’s expeditions. The sketches of wildlife are in the Natural Histry Museum, while we have the sketches and paintings of the places themselves.
I think it may be shortsighted to think that publishing a feed URL is just for humans. Sure, it lets people subscribe in feed readers, but I think it also makes the feed discoverable by people who are writing applications that consume RSS.
Hi Jim
I totally agree. As I said, it was made for machines two years ago. You’ll spot a RSS icon shortly . . . . we also have a geo-Opensearch feed available which needs a quick and dirty bounding box interface built for it (again, probably next week). It outputs objects with GeoRSS within a defined lat/long box or radius.
The Cook travels would be excellent. Great idea.
Seb
Well, after MW2008 I can say that implementing OpenSearch at the Walker has definitely moved up my priority list – especially following a comment on our blog about embedding the DC metadata in the RSS feed. We’ve got the DC bit done, the RSS and OpenSearch should be fairly straightfoward. (he said optimistically… :)
Hi Seb
Good post, along the same lines as the post on RSS I did, but at the time I hadn’t made the connection between what I was suggesting and OpenSearch :-)
For benefit of your readers, I’m gathering together a list of museums resources in the “machine accessible” arena plus some other stuff at http://www.mashedmuseum.org.uk. There is also a mailing list and I’m hoping to find the time to knock out some simple guides to mashing, Y! Pipes, etc.
Please feel free to join the list and chuck some stuff about!
Mike
There you go, we’ve made our feed a little more visible to the casual visitor now. It appears when you have made a search that returns a result (but obviously not when it doesn’t).
We will be adding additional fields to our feed in due course as well as a few nice surprises (of course!).
Now start doing something cool with it!
Have you thought about adding OpenSearch-Geo and -Time extensions to your interface? And also including GeoRSS in your feeds marking the origins of the collections?
Hello, what would be the possibility of this idea, to make a connection between two of your suggestions: 1) Use WikiCharts to get an idea of all the most popular topic searches mapped to collections and then 2) create pages of search results on those most popular topics, allowing any collection organisation to feed search results to it? If you created a facility based on this idea, what would be the possibility of enabling any users to create and share their own search pages on more niche topics? What other possibilities could there be? Could you also build layers into it to connect events, exhibitions and learning resources to those topics?
Andrew – we are about to launch a geo-search but until that is live we are holding back on releasing geo-data in the feed. There are a number of inconsistencies around geo-data as I have blogged about earlier but also with our collection many objects have multiple locations attached to them (where they were designed often differs from where they were made, used etc)
Bridget – wait a few months . . . .