For the Love of Mathematics

I did not intend to take the weekend off. It just happened. And it's not that I was goofing off. I was lost in mathematics.

If I have one regret about my education, it is that I did not study more mathematics. And by that, I do not mean the boring wrote mathematics that the word conjures up. Instead I mean the fascinating kind of math I learned in Larry Moss's Q520 -- Math & Logic of Cognitive Science -- at Indiana University.

One of the tools we explored in that class was Latent Semantic Analysis. It's a pretty fascinating tool that uses the statistical regularities within texts to extract meaning. So Saturday was lost to writing the code in MATLAB to run LSA (which is an application of Singular Value Decomposition of a matrix). And most of Sunday was lost to playing with the program I had written.

I grabbed 6 stories about Iraq and 6 stories about the NCAA tournament and played with the LSA algorithm. LSA can be used to cluster concepts, and this did an admirable job. Concepts such as "Baghdad," "soldier," and "killed" clustered together.

And the algorithm was able to sort these concepts from the basketball concepts such as "Duke," "Blue," "Devils," and "rebound."

There were not enough stories for the algorithm to do anything really cool, but I kept thinking I was on the verge of something cool, and I would write about that. It needs more time, and there are too many other projects demanding my attention right now.

But I encourage you to read up on LSA when you get a chance. It's pretty cool. If I win the lottery, I am going back to Indiana for another Ph.D. in mathematics of cognitive science. (Note the irony that the lottery has been called "a tax on people who are bad at math").