Friday, June 24, 2005

OUseful Info

It feels like an age since I posted here, and checking out the dates of the previous posting suggests the same... The main reason (or one of them) has been time spent elsewhere - for example on my "institutional blog": OUseful Info.

OUseful Info tracks my thinking on matters and hacks relating to the Open University, and in particular (at the moment) tools geared towards the OU Library.

Recent postings there include a few thoughts on library toolbars, and a doodle on why e-learning's not just about web pages (more on that soon...).

Monday, June 06, 2005

Towards the Skeletal Web

The blogosphere went a little crazy last week with the announcement of the Google Sitemaps initiative, but in the true spirit of scholarly research I thought I'd jot down my own thoughts before seeing what everyone else has written...

My immediate impression, and one that I fired off to a couple of OU conferences the day the initiative was released, was that this seemed to me like Google trying to get webmasters to do a lot of the grunt work wrt site mapping. (For readers who haven't come across the thread before now Google are proposing that webmasters publish an XML site map somewhere obvious, and upload a copy to Google).

There's nothing really new about this, I guess - after all, Amazon has long been using the co-developer/user community to build Amazon storefronts with the Amazon e-commerce API...

The second thing that came to mind was - will all my separate Blogger pages now be described in a blog sitemap? (That would be handy...)

The third thought that struck me was what this might mean in terms of republishing the web...

Let me explain: one way of categorising websites is to look at them as: 1) content pages, 2) presented with some styling/branding (and extraneous advertising!), 3) organised according to the site architecture.

It's easy enough to lose the style gubbins by simply republishing each page as unstyle XML, or even RSS, but there is then a problem for the user wishing to navigate your site (other than by maintaining a sitewide RSS link'n'highlights feed). However, if the sitemap becomes widely used, as it may with the weight of Google behind it, and probably will if the other search engines get into it too, then it will provide a way of navigating the feed web easily.

So imagine - I stumble across a web page (a real web page one day, that I like the content of, and that also suggests a good site behind it. Considerate as the designer is, the page is <link /> tagged with references to the RSS feed for the page (in a standardised form easily derived from the URI of the (X)HTML page) and also to the sitemap (e.g. <link rel="sitemap" href="http://wherever.com/sitemap.xml" />)...

...and now the site is mine...I can provide my own navigation, style the content pages as I wish, and dump the ads and extra pages...

The scene is now set for the fourth thought - suppose that del.icio.us+ gets on the case, and people start tagging up site maps. Rather than be forced to navigate the site via the sitemap hierarchy, perhaps new structures will emerge based on individual interpretations of tag sets and clusters. Could it be that site remixes will become more popular than the sites themselves (as as historically been the case with accessible versions of complex, media rich sites?

It will be fun to see...

Del.icio.us tags:

Wednesday, June 01, 2005

Using RSS for Data Interrogation

In my previous post on RSS <textinput /> search queries I recalled an even earlier posting on how it would be cool to have an RSS search facility on Firefox.

My occasional searches for such an extension have to date been futile, which got me wondering about whether the functionality is already there? That is, can I subscribe to an OpenSearch query feed, for example, in Firefox as it stands, or using one of the RSS reader extensions?

So, just to provide a bit of context, here's an RSS returning search on Jon Udell's Infoworld site taken from A9.com:

http://a9.com/-/opensearch/search/B0007WF74K/screencast?count=10&startPage=1

Just by having this feed abvailable, do we get any emergent functionality for free by plugging it into other stuff that's laying around (a bit like network effects Jon himself observed were possible).

My first thought was the Sage extension, which does allow queries into a couple of sites that already provide an RSS search. However, outside of hackingthe XPI package, there didn't seem to be an obvious way to add another feed.

Second up was the possibility of adding a particular OpenSearch query to the Firefox search tool. This is normally easy enough (for example, here's how I added the an OU Library search to Firefox), but from my limited understanding of it requires searches of the form http://whatever.long/path/name?q1=searchterm1&q2=searchterm2 etc. Without a simple URL Rewrite (which I don't have access to from where I'm writing now) I couldn't see how to fit this into the Firefox search extension definition.

Thirdly, and this seems a bit of a cop-out (though it does work), was simply to use a keyword search. The format of the search I'll use will be udell %s, which will search infoworld.com using the search terms captured in the %s variable. Rather than just display the search results as raw RSS, I'll shove them through an RSS2HTML service. If you'd like to try it yourself, here's the keyword search bookmarklet. Alternatively, here's a quick demo of a search on firefox search.

This approach would be muuch simplified (and generalisable) if the OpenSearch definition file (or the A9 ID of the feed) were autodiscoverable. For example, something like this should do the trick for the InfoWorld search: <link rel="opensearch@A9" type="application/rss+xml" href="http://a9.com/-/opensearch/search/B0007WF74K/" title="Infoworld A9 OpenSearch Feed" />.

Del.icio.us tags:

RSS: As Good for Search as for Syndication?

Having just taken receipt of Beginning RSS and Atom Programming I thought it was about time I actaully scanned over the RSS and Atom definitions again...

One optional tag I had completely forgotten about(if indeed I ever knoew about it...) in the RSS 0.9x and 2.0 definitions is the textinput element. I guess this is partly because all the feed generators I reuse don't bother with it...

Using RSS for syndicating content is one thing, but I find more and more that I am consuming RSS in all sorts of ways, oftentimes preferring the RSS-Web to the HTML web. This is partly due to the fact that it's the information I'm invariably after rather than website advertising.

Discovering the information I want is, of course, one of the major issues - this surely is where the <textinput /> tag comes in... It could very easily be used to reveal the search function on whatever is generating the RSS feed I'm reading.

As far as blogs go, I often keep referring to the same few, well-heeled ones, as well searching blogs in general via the Feedster column on my A9 OpenSearch panel.

So- what I shall look at doing with my RSS feeds, is look at ways of inclduing a search link, via a <textinput /> tag, that produces results in an OpenSearch RSS format.

How would this change the look of my aggregator? If my aggregator allows me to inspect the contents of a single feed, it would make sense for a single search box to be reconciled with the RSS search feed for that particular feed. Where my aggregator is displaying information from several feeds at once, the search box could perhaps fire off a search request to all of them at the same time, and aggregate the results.

There is a lot of potential, I think, for syndicating search information for a feed, well as just the most recent content elements. The potential for a Firefox OpenSearch extension is still also there...


Update
Since originally posting this note, I 've read a couple of other posts today that mention the <textinput /> tag. For example, this post on Russell Beattie's blog talks about using the <textinput /> tag to capture comment feedback. If you subscribe to the feed from his blog in e.g. Bloglines, tyou can even see it in action.

Del.icio.us tags: