Over on Burningbird, Shelley has written a great summary/analysis of the current thinking on Technorati’s tags. It is beautifully written, sports some wonderful photographs, and is well worth reading. I’m not even going to attempt to summarise it here, because to do so would be to be like reinventing the wheel in triangular shape – pointless and nowhere near as good as the original.

The thoughts that follow are an elaboration of the comment I left on Shelley’s post, so if you read that then some of this may seem eerily familiar.

As I said on my previous post about Technorati tags, I can’t help feeling that we’re really only at the very beginning of the creation of meaningful tagsonomies and tagsonomical tools. Technorati’s implementation of tags is one step on a long road, but until we can sort by what Technorati calls ‘authority’ (but which is really a sort of popularity), pull the search results in to our aggregators by RSS, search using Boolean operands on multiple tags and do all sorts of complicated bespoke filtering, tags will remain a bit of a kludge.

Tags are, at the moment, at the ‘sledgehammer to crack a walnut’ stage, and there’s a lot of work to be done before we get it refined down to the toffee hammer stage.

A big issue is obviously implementation. People are lazy – I certainly am and I am sure I am not alone. Until we have a way to automatically tag or create tag suggestions that can be approved or disapproved by the user, we are going to have to rely on people bothering to tag their posts, and we’re going to have to put up with the way that the variable quality of their metadata affects this metadata-reliant system.

Of course, we have movement in that direction in terms of the various tagging tools which have sprung up with impressive rapidity. Ecto supports tags using the Custom Tag facility – just create a custom HTML tag with the code below and it will automatically create a tag from the selected text.

<a href="http://technorati.com/tag/%*" rel="tag"></a>

Stephanie Booth has created a plug-in for WordPress, and there is of course the Oddiophile bookmarklet I have mentioned previously. All good starts, but they still require the blogger to bother using them and think clearly about which tags are relevant. As Shelley and others have noted, people are not necessarily very good at creating accurate tags – even people knowledgeable in the area of taxonomy and metadata don’t always create good tags for their own work.

That said, I think there are a few uses for which tags, even as they stand, beat every other system hands down, and one of those is classifying posts by language. At the moment, there really isn’t a consistent way to mark blogs or blog posts by language and that makes it very difficult if one is interested in finding blogs in a given tongue.

If I wanted to find blogs written in Welsh, then I have a bit of a challenge ahead of me. I can search in Google for ‘blog cymraeg’ but all that gives me are blog posts which use the word ‘cymraeg’, so if the post is in Welsh but doesn’t mention the word ‘cymraeg’ it’s not going to show up. For more popular languages, I can choose which language Google should search in, but that still means I need to pick some keywords to search on.

There is a similar problem even with specialised blog search engines, including the keyword search on Technorati – they all search content. I’m no metadata expert, but I see a clear difference between metadata that describes the contents of a post, i.e. what it is about, and metadata that describes the format of the post, such as what language it is in.

By allowing people to add format metadata, tags give bloggers the power to describe aspects of their posts that would not be accurately reflected by keywords selected from the content. Tagging all Welsh posts with ‘Cymraeg‘, for example, allows anyone interested in Welsh blogging to locate the most recent posts in that language, regardless of what those posts might be about.

Using tags to make up for this shortfall in existing blog metadata, we can then use Technorati as an engine for discovery (as opposed to search) within a set of given criteria. At the moment there is just no other way to do this.

Tags may be a bit kludgy at the moment, but because they are capable of filling a gap in the way we locate blog posts that may be of interest, I think they are going to be with us for the long haul.

tag, tags, tagsonomy, tagsonomies, taxonomy, taxonomies, metadata, technorati

Technorati really have been busy recently. As well as the new features I mentioned last week, they have now introduce a new tag search facility which allows you to search for posts that have been marked with a Technorati tag:

<a href="http://technorati.com/tag/[tagname]" rel="tag">[tagname]</a>

For example:

<a href="http://technorati.com/tag/suw" rel="tag">suw</a>

Technorati then pulls in any blog posts which have been thus tagged or categorised (it treats the categories as tags), along with photos from Flickr and bookmarks from del.icio.us, and collates them on a single page.

The link text does not have to be the same as the tag and could even be made invisible by linking to a space, although having discussed the matter with Kevin Marks, who clears up a couple of questions about tags on his blog, there are good reasons why that is not a good idea.

Firstly, linking to the same text as the tag gives people a link straight to the relevant tag page so that they can see who else is using that given tag, a bit like having the Technorati cosmos link on each post. It gives people clues about how you are tagging your posts, which may in turn encourage them to use similar tags so that your tag search results overlap, e.g. if two people are talking about tagsonomies then by tagging their posts as such, they both show on the same page and become part of the same discussion.

As Stephanie Booth points out, this is a very similar effect to that of TopicExchange, which I have used on occasion and found to be useful. TopicExchange works by collating trackback pings so is easier to use than Technorati tags, but Technorati probably has more reach and a better search facility. However, TopicExchange is run on a central server, whereas the Technorati tags are not only distributed, they are also open to being spidered, freeing up the tag data for other services to use.

Another reason for keeping your tag URI and your tag link text the same is that of spam. It’s fairly obvious that the Technorati system is open to abuse, as pointed out by Nootropic, but they are already aware of that possibility. Kevin Marks is very experienced at identifying spam, and is looking at ways to keep the database clean – one way of spotting tag spam would be if the link text did not match the tag.

All in all, this is an interesting way of using emergent tagsonomies to pull together diverse datastreams in one place. As it happens, I’ve had a number of different conversations recently with friends about such things, and this is a useful first step along the way to creating a single entry point for a variety of sources.

I do, as usual, foresee a couple of small problems. The Technorati tags at the moment need to be inserted manually in the markup by the blogger and, as anyone who has worked with taxonomies and tagsonomies knows, laziness wins every time – many people have a tendency to simply not bother adding metadata, no matter how useful it is. You only have to randomly browse the Flickr to see that not only do a fair proportion of users not tag their photos, many don’t even add meaningful titles.

A way round this would be for the blog tools to provide a specific ‘tag’ field for people to add in their metadata more easily, although it would be better if the software could somehow suggest tags for you. That is, though, getting in to the realms of metadata autodiscovery and automatic classification, which is far harder problem to solve than it first appears to be.

Perhaps another option would be for Technorati to use its keyword search facility as well as the tags, so my ‘Suw‘ page would show the most recent tagged blog posts and a selection of the most recent ‘Suw’ keyword search results. I would also like to see Technorati pull in other data streams, such as Furl bookmarks, (not everyone uses del.icio.us), or 43Things tags.

Another problem is one which bedevils all tagsonomies and that is the issue of synonyms and plurals: Do you tag using ‘tagsonomy’ or ‘tagsonomies’? Perhaps Technorati could use a Google Suggest style solution to this, so when you search for ‘tagsonomy’ it asks you if you would like to also see the results for ‘tagsonomies’ as well. It would also be good if one could use Boolean search operands in the tag search facility so that you can create complex search strings rather than being limited to one word or phrase.

Overall, Technorati tags are going down rather well, and I am looking forward to seeing how this project develops. I’ll even try to remember to tag my own posts properly.

Technorati tags: tags, tagsonomy, tagsonomies, technorati, taxonomy, taxonomies, suw