Tuesday, October 19, 2010

Fractonomy and Mandelbrot


When you go searching for a Seth Godin book in the Library, you go to the Business Section and then look for the Marketing shelf. Not only that, you could find Philip Kotler's book in the same shelf.

Taxonomy or the science of classification, makes this happen.

When you go searching for a "london eye" picture in Flickr, all you do is type "london eye" in the search box and thousands of 'relevant' pictures appear. Folksonomy or tags by folks make it possible. The fact that you could also find Big Ben, is not because of an existence of a hierarchical classification scheme, but due to statistical clustering.

Instead of you searching for information, if relevance has to find you, in this exciting period of the "connected consumption", what would be the key enabler?

Fractonomy. It is a portmanteau of Fractal and Taxonomy.

We could use the power of statistical clustering, not only to cluster consumers with 'similar interests. We could also use statistical clustering to map the trajectory of future interests,which would emerge like a fractal.

RIP Mandelbrot.

Sunday, October 10, 2010

Curation without Curators


There are no cars that drive itself yet, no machine that can understand human language, in the marketplace.

But,this week, there was some promising news, about NELL and the Google Car. While this is great news, it might take years, probably decades, to iron out the contextual wrinkles, for widespread use.

Many popular Science Fiction books and Movies, posit a perilous future, with machines taking control over human beings. Such a possibility make books interesting, yet is still a fictional future.
We need imagination to augment the abilities of humans, with the possibilities provided by machines.

Curation is one such application. Statistical machine learning could help solve the clustering problem very well. Having known the cluster that you are interested in, by constructing curator pools, machines can be deployed to do the "hill climbing". ... more like Don Norman writes, in "Design without designers".