This post brought to you via the built-in blog post editor that comes with Flock.

Flock is a new browser built on the Mozilla codebase and I’ve just downloaded a beta version of it. It incorporates a very slick flickr uploader as well, so you can put your pictures up very quickly.

Tabbed browsing as well.

I’ve only been using for 5 mins but I like it a lot already. It’s a nice browser for Web 2.0 types. I’m told it has delicious integration as well. More later.

Blogged with Flock

Returning to my theme of more formal folksonomies. It’s over a year since I wrote that and folksonomies are still very informal.

This makes it hard to find things using automated tools. For example, my depictr toy looks for flickr photos to match keywords in a song. Even when the keyword analysis is good, the pictures returned are of mixed relevance.

RDF would give us the formality we are after, clearly. But the people aren’t going to enter RDF or anything like it in the few seconds they’re prepared to spend tagging.

We want those elements though: subject, predicate and value, from which RDF could be generated. In other words:

1. what is being described (a flickr photo, a link in delicious) - which we know
2. what dimension of the thing is being described (eg: its creator, date of creation, colour) - which isn’t specified in the folksonomies I’ve used
3. the value in that dimension (eg: Mike Harper, 9/6/2006, blue) - which is usually the tag itself

What we get is a word or phrase that has to stand for both dimension and value. For example:

football - meaning topic= football
mike harper - meaning author= mike harper

How to make this more formal? A pragmatic place to start, I think, would be to assume no-one is going to provide more information than they do now when they tag things. What can be done with that?

I’ve been experimenting briefly with Wordnet, a database of words and their semantic relationships. Since in this scenario words are all we are getting, this seems like a useful resource. It can tell us, if it recognises a word, how many senses it has ie: distinct meanings. If it is only known to have one, then we know that the ambiguity of that tag is quite low. If on the other hand it is known to have many meanings, it’s usefulness is limited without further information.

Wordnet also provides hypernyms - words that mean something broader than the word you specify. So for example colour has a hypernym of “visual property". “Visual property” has a hypernym of “property". This might allow some kind of reasoning about the relationship between words, such as whether they represent the same concept.

Perhaps some automated work can be done using resources like Wordnet to try to reassemble some of the meaning intended by the person doing the tagging. I’m going to play with this.

Following on from blogging someone else’s photos from flickr. In that post I was trying out blogging “something” rather than just blogging.

It was a trivial post but I think this highlights two different approaches to blogging. I’ll call them active blogging and reactive blogging. Formulating an idea and writing about it would be active blogging because there isn’t any obvious trigger from outside. Clicking on a “blog this” button or moblogging the scene in front of you would be reactive blogging.

Of course it’s not black and white. No-one has ideas without input from outside. And few people simply react to external events without at least expressing an opinion about them.

But I think it’s a useful distinction because each requires a different frame of mind. I think it also has implications for how a blog is organised.

I’ve got an early version of depictr working here.

This was my idea for a mashup between Yahoo Content Analyzer and Flickr. The idea is to analyse song lyrics and find appropriate pictures from flickr to complement them. Mainly for fun though it might help in some creative activities.

There’s an element of serendipity which is more significant than I’d like it to be at the moment. And there’s lots of stuff to tidy up.

Doesn’t work in Firefox at all at the moment. Sorry.

depictr

Statue Piazza Firenze

Originally uploaded by david.raywood.

Here’s me experimenting again, to see whether you can blog other people’s photos from flickr, which of course you can.

This is a photo taken by a former colleague in Italy.

Dean Clough Sculpture #1

Originally uploaded by mcharper.

I’m just experimenting here with a flickr feature that allows you to blog a photo. You have to give it an “API endpoint", ie: the URL of a server-side script that will receive information and incorporate it into your blog.

You have to specify the username for blog but you can tell flickr either to store your password or to request it from you each time you make a post.

The picture’s not really relevant, I was just experimenting. But for what it’s worth, this is a photo of one of the sculptures on the Dean Clough complex where I work.

I called Martin K’s script to analyse this picture:

Which said the average hue was:

in the XML that was returned:

This could be processed by a script, obviously.

But manually for now, I took that hue, darkened it tonally, and created a mount of that colour like this:

I’ve been on flickr again, my favourite website.

Flickr has a huge number of self-organising groups covering a wide variety of interests. One of the groups I joined is Color Fields. Photos qualify for the Color Fields pool if they are of essentially one colour.

Although I don’t create many images that fit that description, I am interested in this group because of a discussion thread that started when someone suggested tagging images with a code representing the main colour used in an image, so that people could more easily find images of a certain hue.

Then it was suggested that this code might be arrived at using an automated tool and people have been working on these.

I’m interested in this for two reasons:

1. it’s an example of metadata being collected automatically from images. I’m sure this is a rich area of exploration
2. specifically, I want to try to create some online tools that can use image metadata like this, for example to decide what colours might complement / contrast with a photo best, for mounting and framing

Martin Krzywinski has created a script that analyses an image and returns some useful data in XML. I did some experiments tonight using this to analyse some of my flickr pictures.

I set up Skype the other day.

In a nutshell, Skype allows you to phone people on your computer using a headset. It costs nothing if the other party has Skype. Even if they’re on the other side of the world. Excellent.

If the other party doesn’t have Skype, you can still call them at “Skype Out” rate, which for calls from the UK is about 1p a minute. I haven’t tried this feature yet.

I only know one person with Skype at present, a colleague from my previous site. I called him yesterday.

I’ve got a few problems with my voice being echoed back at the moment, which is quite offputting. Apparently most people find a way of “tuning” these problems out so I should be able to sort it out. At one point his voice “daleked” to the point where we couldn’t continue. I’ll have to find out how to address these things.

But despite these problems, I like it. It seems like a new way of communicating. Even though it’s voice, it’s not really like phoning someone, where there’s an established protocol. More on this later.

So to Web 2.0. This has become the focus of my interest.

Some readers might not know what it is so I’ll dedicate this post to that.

Dale Dougherty of O’Reilly Media coined the term during a brainstorming session. The participants were talking about the burgeoning of new applications on the web in stark contrast to the popping of the dotcom “bubble” that had preceded them. These applications had certain attributes in common. They were part of an emerging “Web 2.0″.

For a short summary I recommend this article which boils it down pretty well.

For a more comprehensive description, this excellent article by Tim O’Reilly describes his vision of Web 2.0. There are about 2300 links to this in del.icio.us and it’s become a standard Web 2.0 text.

Being a minimalist I want to identify just one thing from all the features identified in Tim’s article that is the defining characteristic of Web 2.0.

In my opinion that thing is participation. In Web 2.0, people can create and share their writing, photos, knowledge, opinions, experiences, in fact any information they want to.

To me the perfect Web 2.0 application is flickr, the photo sharing site. That’s my model for Web 2.0. I recommend trying it.