Tuesday, March 29, 2005

TinyTagURL via del.icio.us?

Absolutely Del.icio.us - Complete Tool Collection looks like a sensible place to go if you're looking for any del.icio.is tools (you may notice the toolbar link in this blog's sidebar...)

One thing that caught my eye was how a TinyURL is used to capture the link. (TinyURL was first pointed out to me as a comment to this post on shortening URLs, along with the more meaningful notlong.com.)

All well and good - but the meaninglessnes the TinyURL doesn't really endear it to me, though it admittedly much shorter than the URL it redirects to.

What would be far more useful, I think, would be to be able to create a meaningful shortcut. del.icio.us can actually get me quite close, as this link to my del.icio.us delicious tag shows, but it's still a neighbourhood search, rather than a one-to-one redirection link (this is an injective mapping, I think...the other, one-to-many case does not make much sense in terms of mathematical functions, although many-to-one mappings are more reasonable...).

What I'd really like is to be able to generate a del.icio.us.ly simple URI that somehow reflects the tag, or tags, I have used...this could perhaps be as simple as just adding a link number onto the end of the URI - e.g. http://del.icio.us/psychemedia/delicious#nn or http://del.icio.us/psychemedia/delicious/nn. The number (nn) is still arbitrary, but the rest of the URI is meaningful...and even without the actual link number (which may be set using a user preference, perhaps) still gets me close...

Del.icio.us Tags:

Wednesday, March 23, 2005

Who Rates This Page?

As I pondered over several of the bookmarklets I have installed on the browser I use in the office, I realised how my idle-browsing habits are slowly starting to change as I augment my browser with 1-click search tools.

Consider these three bookmarklets, for example: On Technorati?, On del.icio.us?, Google links?. All of them return a search on the URL of the currently loaded page.

On Technorati? and Google links? both see who else has made mention of the page, On del.icio.us? sees who's bookmarked it.

As well as the links embedded in the page I'm looking at, I can also navigate away from it to pages related by backward links, such as you might expect from a Trackback service perhaps (here's a Trackback with Movable Type screencast if you're interested...). Alternatively, I can see who's rated the page enough to bookmark it on del.icio.us - I can then go through the list see how other people have clustered the page, for example through the tags they've used, and I can then go wandering through pages related to the one I was originally looking at through a similiarity measure based on a tag folksonomy.

Tuesday, March 22, 2005

OpenSearch Aggregator for Firefox?

As the viral buzz catches hold of OpenSearch, I have to admit I am quite looking forward to the first OpenSearch Firefox extension that appears on the scene.

There are RSS aggregators already available for Firefox of course (such as Sage) but I think a personal metasearch aggregator would be a bit different...

For example, it may be useful to be able to set up personal search profiles that make queries to different sets of search engines.

To make administering personal search contexts a joy rather than a chore, my v0.1 client would support:

- easy definition of new search contexts;
- one click subscription to an OpenSearch feed;
- easy maintenance/editing of contexts and feeds.

The one click subscirption raises an interesting issue - does OpenSearch support autodiscovery that makes it clear an OpenSearch feed is available? Somehting like this perhaps:
<link rel="opensearch" type="application/rss+xml" href="http://safari.oreilly.com/osrss?search=" title="OpenSearch On Safari" />

Contention/prioritising results is an issue, of course, but I have some thoughts on that which I'll describe in a forthcoming post...

Monday, March 21, 2005

So Who's Open Searchable?

It seems like the release of OpenSearch has appealed to a wide variety of people who run their own searches. Why? Presumably becuase it means that sites now have an easy way of federating their search results and participating in a flexible form of cross-searching.

You only have to look at the growing list of sites listed on A9.com > More Columns to sense that adoption of OpenSearch may be about to grow quickly. OpenSearch columns, by the way, refer to results 'columns' that are available on A9 - that is, they refer to search engines who have made themselves available to A9 through the OpenSearch protocol.

It will be interesting to see whether its lifecyle then follows the rest of the Gartner Hype Cycle:

Gartner hype cycle

It'll also be interesting to see how RSS aggregators evolve to provide customisable, personalised metasearch engines, in which the user subscribes to OpenSearch feeds of their own choosing and over time tunes the sites included to suit their own needs.

Del.icio.us tags:

Friday, March 18, 2005


JISCMAIL is a mailing list service sponsored by the JISC, for the UK Higher and Further Education communities, enabling members to stay in touch and share information by e-mail or via the web.

When you sign up to a list, the listserver can email you each message as it arrives at the server, or batch the mail in a digest on your behalf. The list server also takes care of archiving. I'm particularly keen on a couple of the mailing lists: Creative-robotics-network-info, the mailing list of the Creative Robotics Research Network (CRRN) and psci-com: on the public understanding of science.

Recently, however, I've been suffering an email crisis. I read an article on Managing Incoming E-mail: What Every User Needs to Know at the start of the year in an effort to be told again(!) about effective ways of managing ever increasing loads of email... but it's not working... I dread to think how many messages are in my Mailbox languishing unread and know I really should get round to adding some more filters, reducing the amount of lists I'm subscribed to, or finding another way of consuming the information that doesn't clutter up my space.

So - reading my JISCMAIL lists through an RSS aggregator seemed sensible...but I couldn't find any feeds there...so, here's one for the CRRN list, and here's one for PSCI-COM. The script needs tidying up, of course, but it's a start...

Del.icio.us tags:

Playing With ISBNs...

As with the plethora of services that make use of postcodes, some of which I have blogged about recently, there's also a wealth of services out there (not least the Amazon E-commerce Service) that take an ISBN as an argument.

What caught my eye with this posting - CIE Thoughts: An RSS feed for books? - about isbn.nu was the mention of the RSS output feed option... It's good to see that more people are making their search outputs machine readable...:-)

So for example, how about searching for a book by ISBN? (There's also a keyword search as you might expect...).

Some of my own doodles on using ISBNs can be found on the OUseful course book search page and the not quite ready yet Purchasers of this book may also be interested in studying... page.

Isbn.nu offer an RSS feed of new bools identified by keyword. For example, to keep track of new books on robotics, I subscibe to the http://isbn.nu/xmlsisbn/robotics RSS feed.

Del.icio.us tags:

Thursday, March 17, 2005

del.icio.us Firefox Extension

This nifty little Firefox extension adds a sidebar to your Firefox browser that gives you an overview of your del.icio.us tags, and the bookmarks associated with any particular highlighted tag.

For example:

Firefox del.icio.is sidebar screenshot

Fot the record, I have started developing an as and when tutorial on web development using Firefox.

Del.icio.us tags:

Let's All Search Together - A9 OpenSearch

OpenSearch is a collection of technologies, all built on top of popular open standards, to allow content providers to publish their search results in a format suitable for syndication.

...and very pretty it is too. The idea is for search engines to have the option of returning their results in an extended form of RSS 2.0 referred to as OpenSearch RSS 1.0. To complement this is a standardised query language, OpenSearch Query Syntax 1.0.

The library sector have been thinking about generic queries for some time of course, (for example, check out this Symposium from 2003 on The Next Generation of Access: OpenURL and Metasearch). OpenURL consists of a BaseURL that points to a "link server" capable of resolving the link, plus a query string which contains bibliographic data expressed in a well-defined format [ref: here]

While sort of on the topic of OpenURL, the OpenURL/Google Scholar Firefox extension, made more usable here lets you look up responses from Google Scholar using your local academic library's link resolver... If that doesn't make any sense at all to you, here's an FAQ.

I did try once to get my local academic library to provide an RSS output feed from their Resources for Open University TEachers and Students (ROUTES)...(I offered this as proof of concept, though it's very ropey, doesn't escape special characters, doesn't necessarily give clean RSS etc)...but they preferred a bespoke XML format...

o hum - it will be interesting to see if they like OpenSearch RSS any better...

Tuesday, March 15, 2005

Bookmarking in del.icio.us

Whilst watching a power user at play in the language evolution in del.icio.us screencast, I was intrigued to see how the optional extended element of a del.icio.us bookmark could be populated simply by highlighting a portion of text on the page to be bookmarked...

...but not it seems in the bookmarklet I got from deli.icio.us when I subscribed?

Never mind - the fix is a quick one: e.g. create a new variable:

then add it to the URI used to open the bookmarking window:

Del.icio.us Tags:

Routefinder Bookmarklet

Hopefully this is going to be the last of the recent thread of postcode/bookmarklet posts... Time for pastures new and all that... :-)

...so here we go - a bit of javascript something like the following:
location.href= ''
//targetURI including _from location (homePostCode)
+ encodeURIComponent(window.getSelection())
//the _to location (or v.v)

Pop it into a link and we have the following bookmarklets:

- from myHomePostcode to selectedPostcode (Michelin Routefinder, or GreenFlag Routefinder)

- from selectedPostcode to myHomePostcode (Michelin Routefinder, or GreenFlag Routefinder)

To use the bookmarklets, drag them onto your browser toolbar, set the homePostCode bvariable to your home paostcode, then highlight a postcode (such as MK7 6AA) on an arbitrary webage. Fire off the appropriate bookmarklet to find the route between your home location and the highlighted target location...

Del.icio.us tags:

Monday, March 14, 2005

User Selection Postcode Bookmarklets

Although the use of <link /> elements can be readily exploited using bookmarklets, as the postcode <link /> shows, there are very few that are consistently used and as such even fewer that reach the critical density required to become self-sustaining in even tightly knit communities, let alone across the web as whole.

It therefore makes sense to provide bookmarklets that can feed of user highlighted text. This requires neither any commmitment on the part of the original page designer (no need to ensure the presence of the correct <link /> tag for example), nor sophisticated screen scraping routines to automatically identify information that is not necessarily explicitly marked up.

As an example, consider this description of a Postcode bookmarklet which straightforwardly opens a Multimap map depicting a postcode (such as MK7 6AA) highlghted in an arbitrary web page.

To borrow from that post, here's the
bookmarklet and

As before, it's easy enough to provide a wealth of related bookmarklets, like this
UpMyStreet bookmarklet or this
UK Postcode to Grid Co-ordinates Bookmarklet.

To use any of these bookmarklets, just drag them onto your browser toolbar, highlight a UK postcode on any webpage, and fire off the bookmarklet...

Del.icio.us Tags:

Bookmarklets, Form POST Data and More Postcode Info...

One thing that has been nagging me about bookmarklets has been how to use them to pass information to a webpage/service tht accepts form POST data.

Easy enough with a GET, of course, you just add the variable name and the required data to the URI. But for services like this list of bus/coach services which only accept FORM data, I was stumped as to how to do it...

...until someone suggested making a form... a quick search later, and this page on Smart Bookmarks and Bookmarklets showed the way forward...

My first attempt pulled the postcode <link /> info from a page and then tried to create a new document to hold the form...but that seemed to hang, so now the change is made to the loaded page directly... Naughty, I know, but there we go - if you can show me how to get a working bookmarklet that creates the form in a new document, please add a comment...

So, for the record, this is the script:

if(el[i].getAttribute('rel').indexOf('postcode')!= -1){
f.method ='post';

and this is the bus timetable bookmarklet.

If you want to give it a try, this whole blog is postcode <link /> enabled...

Del.ici.ous tags:

Sunday, March 13, 2005

Web Page Plus: UK Postcodes

A couple of days ago, I blogged about a postcode <link />, implicitly hinting that it might lead to some interesting Web Page Plus functionality through bookmarklets and Firefox extensions...

To move this on a bit, it seems sensible to provide a bookmarklet generator to support the easy creation of links into an arbitrary postcode service... And when I get a chance, that's exactly what I'll do...

Of course, if pages don't include something like <link rel="postcode" type="text/html" title="mk7 6AA" /> the bookmarklets won't be of much use, at least until I work out how to build a Firefox Extension that will let me use reg-exp identified postcodes in an arbitrary page (I know, I know, the ethics of rewriting other people's pages in your browser may or may not be a good thing... see The Google Autolink Debate for a recap).

Anyway, to be going on with, there are a couple more demo bookmarklets... Last time, I demonstrated the postcode <link /> with this
UK Postcode Map Bookmarklet. It's easy enough to tweak this code to access other services by changing a single argument, so here are a few derived bookmarklets...

- UK UpMyStreet Bookmarklet

- UK Postcode to Grid Co-ordinates Bookmarklet

If you customise your bookmarklet with a 'home' variable setting, then it's straightforward to bookmarklet "arity 2" postcode related webpages... You can imagine using this to bookmarklet a route finding service between your home, say, and a postcode <link />ed page, such as these two:

- Michelin Routefinder Bookmarklet

- Green Flag Routefinder Bookmarklet

Del.ici.ous tags:

Thursday, March 10, 2005

seamlessUK - Interoperable e-government?

The seamlessUK website claims to allow citizens the chance to "retrieve information from multiple websites and databases in a single search without the need to know, or enter the specific addresses for each one" via Z39.50 (which I've heard of...) and Z'mbol, which I haven't, though it appears to be something that sits on top of Z39.50, so that's okay then...

XML and SOAP is mentioned in the mix, but any really useful details appear (at the moment at least) to be sadly lacking...

Now, Z39.50 isn't really my area (is anything?) and though I thought I'd blogged about it before, it appears I haven't...

So, as far as my limited understanding goes, I though Z39.50 was old and boring, and XML was to be the way forward with SRW. This reasonably recent post on a Web Services Approach for Search and Retrieve seems to introduce some of the issues (there are a few more links on this earlier posting about SRW, where I suggested that the SOAP/REST debate is going on in the library world using different acronyms...)

Anyway - back to seamlessuk and another quote: "[t]he seamlessUK geocoder will search a number of geographic data sources to resolve place name queries, understand what the user has entered and interpret that query for the different information sources." Good, eh?

Now what gets me is that the taxonomy underlying this system (because you have to have a taxonomy, right?, or ontology, or whatever translator glue you want to call it) is licensable... and there doesn;t appear - at the moment at least - to be an open interface to the system. Which is gonna make it hard for people to embed in their own systems, presumably...so they either won't bother, or they'll build their own...or perhaps they won't...we'll see...

Autodiscovery and Location Aware Web Pages

Autodiscovery describes the way in which machine readable metadata can be extracted - and exploited - from a webpage. The key to it all is the humble <link /> tag, though when the spec. was written things like RSS/Atom autodiscovery were presumably as yet unimagined...

...and so it came to pass that autodiscovery came to prominence with the rise of the blogs, as this early blog posting suggested it might, and this posting shows how it has come to do so...

Autodiscovery is not just limited to RSS/Atom though. FOAF Autodiscovery aims to promote the development of Friend of a Friend (FOAF) social networks by making FOAF records machine discoverable/readable.

The development of web technologies thrives on reuse - and tools that make reuse easy. If you want to generate a bookmarklet to assist you with autodiscovery from a <link /> tag, try this autodiscovery bookmarklet generator. With almost no effort, you can generate a bookmarklet that learn something from a suitably <link /> augmented page...

...which is where the Location Aware Webpage comes in...

The prototype is very, very simple... add a link tag to the head of the page containing the postcode of where I'm at:

<link rel="postcode" type="text/html" title="mk7 6AA" />

(In the US, this would be a zip code, of course...but this is just the first prototype, remember...)

Now for UK Postcode Map bookmarklet straight out of the generator:

if(el[i].getAttribute('rel').indexOf('postcode')!= -1){
var newlocation='http://ouseful.open.ac.uk/postcode/';
var foaf=el[i].getAttribute('title');

This code makes reuse of the http://ouseful.open.ac.uk/postcode/ redirector, which takes a UK postcode as a final argument and redirects it, with suitable GET sugar coating, to Multimap.

So, to see what happens, grab the bookmarklet, place it on your toolbar, then (as long as this page at least is loaded) try it out...hopefully, you'll see where I'm at as I write this page...

So, where next?

A zip code equivalent for the US, a Firefox plugin so there's a little button, like the RSS subscribe button down in the bottom right hand corner, to launch the autodiscovery location map plotter, a user dialogie to choose your locale (and hence your map plotter) and so on...

Any volunteers? ;-)

Wednesday, March 09, 2005

Building Social Networks: FOAF and XFN...

XFN (XHTML Friends Network) and FOAF (Friend Of A Friend) are social networking protocols intended to identify links and relationships between people in a machine readable way.

This comparison of XFN and FOAF (on the XFN site, so maybe there's just a teensy weensy bit of bias, even if it is written by Eric A. Meyer) suggests that XFN offers a far more lightweght approach to the RDF overkill of FOAF. That's not necessaily a good thing, though, as some bloggers have a real problem with it: What does XFN to offer, then, exactly?). For the most considered view I have founfd to date, there's this response to Meyers comparison. However, having only just come across both of these schemes, I shall refrain from judgment until I've had a chance to play with both of them...

On the face of it, though, FOAF seems a better bet. And if there are occasionally useful XFN links out there, the something like this XSLT XHTML/XFN2RDF/FOAF converter seems sensible...

Capturing Personal Network Relationships

Working from a variety of desktop and laptop machines, file, contact, bookmark etc. management can be a real problem if you aren't very discplined as to where you put stuff! Portable data storage is okay for some things, network/web storage for others. So what I'm trying to do is train myself to use web-based facilities, and then glue them together with whatever protocols and XSLT's I can.

Okay, okay, so how off the ball am I? Having only just discovered the del.icio.us social bookmarks manager, I'm keen to explore what is has to offer in terms of bookmark sharing and management. Support for clickable descriptor tags ("one-word descriptors that you can assign to any bookmark"), user-defined views over bookmarks, and the all important REST api suggests the need for some serious play over the next few days...

Friday, March 04, 2005

PocketThis: Why Should I pay for the Privilege?

PocketThis :: Partners Portfolio has several "See how it works" demos of their service. Find the "Pocket This" link on a web page, and get the nuggets of information shown there (travel times, for instance, or classified ad details) sent top your mobile.

All well and good - but you pay for the privilege. Now this is one of those old 'who should pay?' chestnuts, but in this case it may well be that I will use the info pocketed from a company's webpage to make a purchase from that company. In which case, I definitely do NOT want to pay...especially of those self-same companies send me junk txts...

I much prefer the affitilate idea where I get txts for free from a 3rd party, such as my old favourite Txtbux, and they hope i may a purchase that they can claim an affiliate fee from. Of course, txtbux does require registration, which is a big spoiler...

So perhaps there should be a compromise... The current no registration model, where I pay to recieve a txt; or a new model, where I register with Pocket This and get txts for free. If I make a purchase as a result, quoting a code that tracks back to Pocket This, they get a fee, I get a free txt, the company trying to sell me something pays...

Where Am I?

Mobile Commerce, a UK company specialising in content for mobile and location aware clients, describe a Location Gateway that provides location awareness for the four major UK networks (which are unstated...).

For more information, it's worth skimming through their Location Gateway FAQ.