Clustering your own tags

clustering delicious tagsSomeone pointed out this clustering application for Its just for clustering your own data though, so its not leveraging the group mind. Still it was interesting to see the clusters formed by my tags. Some of the clusters were: clusteranalysis(!); visualization; mobile research; market research; flash; ajax; cardsorting; design.

Expectedly, they are using k-means clustering algorithm. I played around with the number of clusters and the cohesiveness of the clusters does change as the number of clusters goes up.

I looked at the mindmap maker as well. Was not able to view my own map (browser had some problems loading Java files. when are they going to be able to create a smooth install experience for java!).

All these clustering efforts lead to one conclusion. I am repeating myself when I say that tags call out for clustering. Watch out for more types of clustering efforts. It would be ironic if someone someone tried hierarchical clustering, and created a personal taxonomy – which people like Clay Shirky have railed against.

5 thoughts on “Clustering your own tags

  1. I am with Clay Shirky here…for the most part. The meaning of “Is” is relative, and categorizing would become straight-jacketing of my (modeller’s) experiences for everybody else. And a user’s perception of “things” change over time too. So those categories that I had made a while ago for myself, may alter or lose meaning to me now. It’s a hard question, categorizing. (eg No flamebait intended, but is christian “science” science or theology?” Well it depends on who you are). And add to that making information palatable or accessible and usable.

    Sorry for a long rant, but I had encountered similar questions while ago on a project of mine, and since then I am trying to understand if categorizing of abstract human experiences (a set of highly personal, subjective “things”) that we call knowledge is possible, and if so, at what level? Or may be there is a totally different way to access this pool of experiences waiting to be explored.

  2. Bhaskar,

    If user’s perceptions change, then either tags or catgeories are likely to beocme outdated. Yes, categorizing is hard, and i think tags hold a lot of promise – especially as a social system for finding.

    However, categories are not going away anytime soon. I believe that because there is a lot of cross cultural evidence that people think in terms of categories. Its something we are good at – cognitively speaking, and its a very efficient way of organizing information. Try to explain the animal kingdom to a child without talking in terms of categories!

    LIke categorization, other methods of filtering also have their biases, and inflexibilities. For example, think of the type of weighted lists that are popular nowadays. Its a very popularity driven way of navigating information. Even within my own data, the most popular tags are most available.

    In answer to your question: categorization of abstract human experiences is definitely possible and happens all the time – cognitively speaking. If you want to build a stable information filtering system, thats another problem.

  3. I am surprised you could not make the mindmap work for you. I am aware that there are some problems when the number of bookmarks are too big. But you seem to have less then 200 bookmarks, so there really should be no problem. Could you specify to me better what the problems were?

    Many thanks,

    P.S. Also the mindmap program creates a hierarchical structure when possible, but as the number of links increase the probability that a tag is compleytely contained in another decreases, and with that the depth of the tree (versus its width)

Comments are closed.