My Photo

« The scoop on CustomScoop | Main | Document Analysis and Blogging »

August 05, 2005


Kevin Burton

If you want more interesting reading about the problem with tags I've been blogging about the subject a lot recently:

Xtags developer

I do not think one (search agent) can get sufficient information from current web page text alone. If I post the lyrics to my song about the trial and tribulations of being born ugly (a song I might call "Frog Pissed") a keyword finder may correlate the webpage with frogs, ponds, amphibians, French, and toilets based on my lyrics. (And, is there a corollary to "Six degrees of Kevin Bacon" which states that all webpages are only all 3 degrees of keyword correlation away from porn?)

The keyword classification might be "fair", but this is one dimensional: perhaps the key is to know that these are lyrics to a song. Certainly that they are lyrics by rapper "Ugly Eric" will change the interpretation of what those words mean.

Catching that classification, a contextual dimension, especially given other subjects and even advertisements on the same webpage (which, if Google Ads, will reinforce the keywords, not the context) could be quite non-trivial without another layer of tagging.

But it is still a ripe field, I think, even if there is more thinking to be done on the subject.


P.S.. I have been mulling over an approach to tagging that is a bit different. I think a successful approach has to involve (1) an authoritative repository of tag definitions, which tags are openly created by the web community; (2) web page writers themselves; and (3) web surfers, via a feedback mechanism.


"The systemic result of this is that there are many tags that you need in order to capture everyone's idea of metadata."

just one point: in an open tag-able system, there is no need for an individual to capture everyone's idea. Just capture your very own associations (and if you consider yourself as a system, which is able to draw a relationship to object data from words you assign yourself - then those are metadata) and wait for others to fill in the complementary tags for a decent recallability for the others you might have missed.

Otis Gospodnetic

I think Saurier here makes a good point. Everyone is concerned with all other people's tags. Is that really a concern? Do you really care about everyone else's tags? Do you really care about 20 billion web pages that Yahoo now indexes? Do you really care that your web search resulted in 347,084 matches in 0.2 seconds?

I think the answer to all these is negative. I care mostly about my own information. Then I care about the information and knowledge of people I respect or associate with, my friends, family, colleagues, respected and knowledgeable individuals. The last ring around me is everyone and their tags. I care about them, but only a little. I care about them only because they provide a big pool of various levels and types of knowledge, which allows me to venture out of my circle, find something or someone new and exciting, and make a connection with my inner circle, pull the link from that external entity to something closer to the core.

It is not a coincidence that this mimics the real social life of us humans.

For what I'm talking about - watch Simpy ( ) in the coming months.

The comments to this entry are closed.

Twitter Updates

    follow me on Twitter

    May 2015

    Sun Mon Tue Wed Thu Fri Sat
              1 2
    3 4 5 6 7 8 9
    10 11 12 13 14 15 16
    17 18 19 20 21 22 23
    24 25 26 27 28 29 30


    Blog powered by Typepad