Blog Changes: Taxonomy and Folksonomy

What’s folksonomy, you ask?  Even Microsoft Word is asking; I just added it to my dictionary.  Taxonomy, as of course you know, is experts stuffing stuff into pigeon holes.  Folksonomy is us folk categorizing anything on the web any way we damn well please.  Is it useful?  The jury is still out.

It’s all about finding stuff.  Hierarchical folders are so twentieth century (I blogged about that here).  The Dewey Decimal System? Fuhgetaboutit; we’d have to go to thirty digits.  The world of the web is flat (actually dimensionless) so we need fast ways to search all of it to find the precise content we want.

Almost since the beginning of the web, html pages have had metatags put on them by their authors to identify their content.  Search engines used to look at these but rumor has it that they were terribly abused by those who were trying to clamber to the top of search engine ranking (how many synonyms for sex can you think of?) and that search engines now ignore them and concentrate on visible content.

Many blogs use category tags.  For example, this post is tagged with Weblogs and Nerding because I put it in those categories and the categories became tags.  TypePad which hosts this blog gives me a convenient tool to add more tags to each blog; I plan to add Folksonomy as a tag even though I don’t have it as a category.  I could also use Technorati or other tools to associate more tags with this or any post on a blog which I own.

You can use Technorati to find blog posts which have tags that you are interested in.  Of course you can also use Google to find posts or other web sites which contain words that you are interested in.  In both cases you will tastefully be served ads which correspond (maybe) to your search terms.  Technorati also serves up pictures with matching tags from Flikr and Buzznet.

What’s the difference whether you search through tags or keywords? 

If you do a tag search of Technorati for voip (go ahead, try it), at the instant I’m writing you will get back 4,595 posts – actual results will vary.  Technorati will tell you how many other blogs point to the blog containing each post which is supposed to be a rough measure of something but no one is quite sure what.

If you do a Google search for voip, you get back “about 27,200,000 … with Safesearch on.”  Google uses a mysterious algorithm which every search engine optimization expert claims to understand to decide which to present first.  Technorati is only searching blogs and Google is searching the whole web so this is not quite a fair comparison.

If you are on blog.tomevslin.com and not in a blog reader, you can click on the Technorati logo in the left column to get an uptodate rank for this blog in terms of inbound links (currently ranked 9,001) plus a list of the blogs which point here and how many inbound pointers each of them has.  Many more readers come to Fractals of Change from Google than Technorati.  But there are more Google than Technorati users and I have no way to measure whether either class is finding what they really want.  Om Malik posted today about adding more tags to his posts and I’m going to do the same and see how useful this is and whether it increases traffic. Comments are welcome.

Now to folksonomy and one of the changes I made to my blog.  del.icio.us is a web site which allows YOU to tag any URL with any tag you want.  (Full disclosure: we are indirect investors in del.icio.us).  There are two reasons why you might want to tag (three if you count whatever motivates graffiti artists to tag).

Reader Abby who spotted the “Tag with del.icio.us” link at the bottom of the post points out in a comment that the del.icio.us user interface is not yet delicious or intuitive.  I agree and I think that’s being worked on.   Meanwhile I’ll give some simple instruction for using it and avoiding much of the user interface.

del.icio.us calls tagging “book marking” because it IS a very good way to mark stuff you read so you yourself can find it later.  If you are interested in solar energy, for example, and you found my post on that interesting and YOU tag it with “solar”, this is much more useful to you for finding it later than my putting it in the category “solar”.  Whenever you want to see the documents which YOU have already found relevant to your interest in solar, you navigate your browser to del.icio.us/yourusername/solar and get links to ONLY the documents you tagged.  You can give a single document as many different tags as you like and later find it again according to any of those tags.

The second reason for user tagging is to help you find interesting web pages you haven’t seen before, same thing a search engine does.  If you navigate your browser to del.icio.us/tag/solar, you will find all links which ANYONE tagged solar using del.icio.us.  The question is does this add value when the article may already have been categorized by the author as solar and probably contains the word “solar” so it would be found by search engines.  Fred Wilson argues that it does here.  Om Malik is more skeptical here.

If many people give a link a particular tag, does that mean it is relevant to that tag?  Will users tag differently than authors do?  If so, is that helpful or random?

I think that user tagging MAY turn out to be useful in ways we haven’t thought of yet.  Look how useful a form of tags are in creating online reputations on eBay, for example.

I’ve made it easy for you to tag posts in this blog to see whether you do it or not, how you tag them, and if that tagging gets new readers.  I’ll let you know what I find out.  Please let me know what you think.

Comments

[David Gibbons]

... if you do get into delicious ... here's a howto I wrote that's a hack for using del.icio.us tags to automate editing a link-roll on your blog's sidebar with 3 clicks while browsing.

http://poductivity.blogspot.com/2005/08/blogging-101-3-click-blogging-with.html

Should be useful to anyone using blogger - but also anyone tired of cutting & pasting links.

[Tom Evslin]

Otis:

Thank you for commenting. I'm happy to have you give this full description of Simpy here. At first glance it looks like you should be a proud father.

Tom

[Noel Welsh]

Though useful, I think we need to look beyond searching and tags. There are many types of information need. For example, sometimes you want stuff on a specific topic. Here searching and tags are pretty good. Sometimes you want that page you visited yesterday. Here your tags and your browser history are good. Sometimes you want to know the most important pages that are related the one you're viewing. Here you want some cluster analysis. No doubt there are more types of need; these are just the ones that came to mind. The current systems go some way to answers some of these questions, but there is nothing I'm aware of that answers all of them.

[will]

Here something I wrote that examined tagging from the perspective of e-commerce.

http://hitchhiker.blogsome.com/2005/07/01/on-ontology-tagging-seach-commerce/

Love to get your take on the ebay reputation = tagging phenomenon.

[Abby]

Way to flatter a girl by linking to her comment. :)

I thought clicked through and thought about registering, but then I didn't. I couldn't decide whether I wanted to use a pseudonym, and then I wondered whether I would get tired of my login.

I think you can easily run into the Kerry problem, the pheonomenon where people voted for Kerry, because they thought that he was electable. Should I use a tag that makes sense to other people or one that makes sense to me? And will I remember the tag that made sense to me at 2 in the morning?

I think that there are two areas where it's really useful.

1.) I can see what my friends and family think is interesting.

2.) Friends and family can create tags of things that they think I might like. Brad Feld has one of these. Sorting this could get complicated, but it's more fun than getting lots of e-mails. It's less useful when you're talking about a famous person or even your Congressman.

[Otis]

Here is something for everyone.

For Tom: http://wiki.osafoundation.org/bin/view/Journal/HierarchyVersusFacetsVersusTags

This is a good page from an OSAF about hierarchies and tags (facets). Pros/Cons of each are mentioned, along with visual UI components as examples.

For Adam: search + tagging you say?
Have you seen Simpy? http://www.simpy.com/

Search + tagging is one of the principal advantages of Simpy over, say, delicious (sorry to post this on Tom's blog).
Guess how I found the link I posted above? I entered "mimi" into Simpy. Guess how many results that returned? 1! Did I use "mimi" as a tag? Nope. But I did remember that person's name, and I knew that I don't use "mimi" much, so the search was very precise, yielding exacly what I wanted to recall.

So, yes, search + tagging. :)

For Noel: so search + tagging + powerful search operators for narrowing and expanding queries?

Again, Simpy to the rescue.
Example:
http://www.simpy.com/simpy/User.do?username=otis&q=tagging

Note the + - and ~ "operators" next to related tags on the right. Click them. Click one, two three, any number you want. Require tags, prohibit them, or just add them to your query. Forget "tag intersection" business as delicious calls it. It's a Boolean AND query. Simpy's comes with stemming, too, so we can put the "but what about plural versus singular form" discussion to rest.

For all:
If you want to play with things I described, you can use a read-only account: demo / demo (username / password). Have fun.

Disclaimer: Ich bin Vater Simpys. ;)

[Adam DuVander]

Perhaps a combination of searching and tag-browsing would solve the problem Noel mentions?

[Noel Welsh]

I'm rather sceptical about tagging. The problem is it just doesn't scale. A search returning, say, 20'000 pages tagged 'solar' is no more helpful than getting back 20'000'000 pages from Google. At least with Google you can add more search terms to get better accuracy. Tags tend to be fairly broad, so adding more probably won't improve the search. Add to this that your 'solar' might not mean what mine does and you've got a real problem. I've grizzled about this more on my blog: http://www.untyped.com/untyping/archives/2005/07/tags_and_the_po_1.html

[fred]

tom

i just clicked through and tagged this post in delicious

maybe you'll end up on the popular page!

fred