RSS search results

A quickie (as I’ve only got a week to go until Museums and the Web and I have workshop on blogging, a workshop on mashups, a professional forum on “openness” and a “blogathon” to prepare…)

but…

I’ve been playing about with Yahoo! Pipes a fair bit this week and preparing some stuff for the mashup workshop. In doing so, an idea I’ve floated before (see slide 33 “provide alternative routes” on the presentation below) came up to the surface again with a very, very simple suggestion:

All museums (everyone, actually) should provide their search results as RSS

Now I can hear some people at the back shuffling around uncomfortably and muttering things like “shoehorning technology”, “RDF”, “Z39-50″…I have my anti-makelifemoredifficult earplugs in, though, and I can’t really think of any practical reasons why this isn’t a hugely good idea.

Searching the web for things like “museum search RSS” doesn’t get me anything useful. From a previous bookmark I had a link to the AADL catalogue – they do it and here’s a search for “Montreal” delivered via RSS.

I’m immediately able to mash this up using Pipes, hack about with the URL, style it using my own XSLT.

Assuming this isn’t an original idea (and Owen Stephens seemed to think it was a good one and had been done within some library systems) – why aren’t more institutions doing it?

Comments please!

[slideshare id=176658&doc=web2-and-distributed-services-mike-ellis-v2-1195806210354162-3&w=425]

20 thoughts on “RSS search results”

  1. I don’t know if you’re familiar with Drupal (www.drupal.org), but its worth a proper look, particularly its Taxonomy system – every category term or combination of terms can generates its own RSS feed by default, providing an excellent base for advanced web2.0 functionality (like you’ve mentioned above) – I’ve often thought there should be a museum distribution of Drupal, its perfectly suited to the task, and seems way ahead of other open source projects of a similar type….

    Reply
  2. Hey Tom

    Cheers, I’ve done a bit with Drupal. It seems pretty cool and I believe it powers the http://conference.archimuse.com/ site for Museums and the Web – currently on my mind as you’ll see from my post 🙂

    I guess the question – and to Seb as well (cheers Seb!) is more about *why* this isn’t more widespread as a generic technique: yes, I know the technology is there and can do it, but why aren’t more institutions -actually- doing it? Is this *purely* a communication issue between techs and non-techs? Or are there other reasons for not?

    Mike

    Reply
  3. Hi Mike

    It is just (again) an argument between the forces of ‘lets have the best’ versus the ‘lets have something we can have now that is adequate’. Opensearch falls into the latter category.

    BTW we just made a RSS search result by latitude/longitude live today also . . . . will show you in Montreal.

    Seb

    Reply
  4. I just knocked together a quick RSS output for the NMM collections search – just add ‘format=rss’ to the end of a search URL eg.
    http://www.nmm.ac.uk/collections/requestHandlers/doQuickSearch.cfm?searchterm=london&format=rss

    It hasn’t been tested at all – it’s literally five minutes old – so it tends to give XML parse errors from invalid characters in the feed. Mind you, the AADL search for ‘Montreal’ also gives XML parse errors when I try putting it through an RSS reader. I think that might be the barrier to putting out text from collections databases as RSS – XML is tricky to do properly and easy to break just by typing punctuation in an object description.

    Reply
  5. Hi Mike,
    Though not technically a museum site (but based within the British Museum), you can get rss results from the Celtic Coin Index (http://www.finds.org.uk/CCI/), all searches can return rss or geoRSS and if you have login rights you can get JSON, XML, KML or CSV. I’ve also tried to build an api for people to play around, but I didn’t have any funding to do everything I wanted. There’s some interesting reuse of the data pulled from this within Sean Gillies’ Mush application and displayed in Google maps here – http://tinyurl.com/yqalcp
    We introduced RSS search results from the Scheme’s database (www.findsdatabase.org.uk) in 2004 and many researchers use the facility to alert themselves to when a new object has been recorded that meets their parameters. For example a silver denarius of Hadrian found on the Isle of Wight.
    Dan

    Reply
  6. Hey Dan

    That’s interesting – I came across the Celtic Coin Index while I was looking around for museumy RSS stuff, and bookmarked it for some later poking about…

    Finds looks like it could also be cool (although I might just have broken it – search doesn’t seem to work…). I’ll check it out another time maybe

    Thanks for getting in touch

    Mike

    Reply
  7. Jim O’Donnell just emailed me:

    Hi Mike,

    I tried commenting on your blog, but my comment seems to have been eaten by wordpress.

    As of this morning, one of our searches supports RSS by adding ‘format=rss’ to the URL: http://www.nmm.ac.uk/collections/requestHandlers/doQuickSearch.cfm?searchterm=london&format=rss

    That took about 15 minutes to put together, based around the same code as our secret hidden JSON feeds:

    http://eatyourgreens.org.uk/archives/2007/11/how_to_make_aja.html

    I knocked it together very very quickly, so it’s not tested and may throw XML parsing errors if the output contains non-ASCII characters. I noticed that’s true of the AADL catalogue too – the accented e’s in Montreal and Quebec can screw up an XML document if not encoded properly. I think that’s the reason this doesn’t get done more often. RSS is easy to publish, but well-formed XML is tricky, particularly when your content is a bunch of text typed into a database by people who really shouldn’t have to know about character encodings and character entities and all that XML crap.

    Cheers

    Jim

    Reply
  8. RE: “*why* this isn’t more widespread as a generic technique”.
    Possibly because I don’t think the museums world has time to find out about all the new-wiz-bang technologies (unless you are a larger org or an electronicmuseum;). And there doesn’t seem to be (and I may be wrong here) an overriding collaborative effort to share relevant technologies. Shouldn’t the efforts of larger organisations be filtering down to the smaller ones?

    Generally its very good to see examples of how things actually work in the real world, and then for tools to be provided so those examples can be repeated be those less technical – does this currently happen in museum land? (& I’m talking as an ousider here).

    Reply
  9. The Bolton Museum web site is based on the Plone CMS and generates RSS feeds for searches.

    In fact this functionality is used in tandem with Feedburner to generate dynamic events feeds for internal pages (it would have been expensive to custom build this into the CMS).

    Does this count as an example of a “smaller museum” filtering new technologies upwards for a change?

    Reply
  10. Thank you Mike for supplying the link that I should have in my first comment!

    Mike’s example from the Bolton site is for site-wide searches – in the next month or two we are will be adding a collections database.

    This won’t be particularly groundbreaking in itself (it will be a basic keyword search, thumbnail galleries etc.) but as I’ve mentioned above you can easily add cut and paste dynamic content to your pages using RSS, a bit of html/css and services like Feedburner. So there may be applications for the resulting feeds that I haven’t thought of yet.

    Examples can be found on any of the “What’s on” pages on the Bolton site; e.g.

    http://www.boltonmuseums.org.uk/whatson/facetofaceexhibition17may

    In this example the generated content can be seen below the Google map on the right-hand side.

    If you scroll down to the bottom of that page you will see “feed keywords” which is my DIY tagging system.

    After adding keywords to selected pages I did a site-wide search for them, copied the resulting feed into a Feedburner account and using a feature called Buzzboost generated some code that can be pasted onto any page relevant to that feed.

    This method also works for external services: there is a Bolton-centric feed from the Portable Antiquities Service on the following page:

    http://www.boltonmuseums.org.uk/whatson/archaeological-finds17apr

    Probably a bit too primitive to count as a mash-up but hopefully this will give people an idea about the potential of having search feeds on their sites.

    Perry

    Reply
  11. Hi All,

    Yes, search results as Atom are a nice option, and I agree with one of the earlier commentators about doing this with Open Search. Here’s an example test implementation from Open Context (not really focused on museums, but sharing field notes / items from archaeological digs):

    http://www.opencontext.org/database/open_search.php?q=copper

    Just change the query parameter and get an Atom feed based on your search terms. This is similar to the example Perry gave for the Portable Antiquities Service.

    Best!
    -Eric

    Reply
  12. Mike–

    Omeka also has RSS feeds for search results out of the box, which could be another option for smaller institutions. Just add &output=rss2 to the query string of the search results URL (e.g. http://example.com/items/browse?search=foobar&output=rss2). It also works with any of the search criteria, so you can do this with a more complex search. For instance http://example.com/items/browse?type=Document&tag=gorbachev,stalin&output=rss2 would give you the RSS feed for all Documents that were tagged both gorbachev and stalin. Again pretty similar to Perry and Eric’s examples. We’re also talking about implementing ATOM.

    Tom

    Reply
  13. The main reason I was never able to get it done was the huge wall of indifference within the institution. I was able to handcraft one for the Dino Directory (but that’s a canned search). I suppose that could be interpreted as me not being able to sell the idea internally well enough. But this was a couple or 3 years ago so perhaps I was jumping the gun.

    Reply

Leave a Reply to Mike L Cancel reply