Computers and I, or computers and me?

I’m always in search of that grand life direction that will synthesis all my interests, values, and subjects I’ve studied. Last year I discovered a field that bridges my fondness for linguistics and computers: Computational Linguistics. Its applications range from automated over-the-phone tech support (bleh), artificial intelligence stuff (meh), or smarter search engines — i.e. ways of rummaging through our modern swamp of information (hmmm). I think there is much yet to be accomplished in the design of intuitive learning tools. My goal is to get people away from traditional computer monitor and keyboard. I’d love to give someone a gadget that they could take into the woods and speak questions to about the trees or soil types. With that dream in mind, I have been working on some infant programming projects that dottle in the basics of computational linguistics, or “natural language processing”.

tagger_screenshot.pngWhen us humans first learn to pick apart English, we find helpful to identify what’s a noun, what’s a verb, and maybe whether a phrase is in the past or present tense. Sometimes English teachers call this part of “mapping a sentence.” For computers that task isn’t so straight forward. A lot of research has been done on the patterns, statistics, and theoretical structure of natural language, just so a computer can determine things like whether the word record refers to a vinyl disk (one of its noun forms) or “the action of capturing sound” (verb form) when used in a sentence. This process is known as “part-of-speech tagging”, and is one of the initial steps a computer program might take towards “understanding” a phrase. While researching the current work in Natural Language Processing I stumbled upon Dr. Yoshimasa Tsuruoka’s part-of-speech tagger that he designed at the University of Tokyo’s Computer Science department. In all my compu-linguistical gittyness, I got inspired to make a graphical interface for his unix command-line program so that non-nerds could use it and get excited too. A little PHP, Javascript, and CSS later, I present The Pretty POS Tagger!


The Order of R

A database-driven, online art gallery. Images, audio, and text “boxes” are displayed dynamically on each page load according to their tags and other metadata. Google image search is by far one of my favorite Internet toys, but I wanted something more meaningful that would allow a user to notice existing relationships, make new connections, and envision whole scenarios from what began as a random pool of media. The idea for the order of R has been in the works for a long while, before tagging was taken for granted as much as it is now. In July 2006 I began learning the Ruby on Rails web framework and my vision has been able to take shape more easily than I ever thought possible. If you visit today the objects are still tagless, there are no audio boxes, and the domain seems more like a blog portal. Well, yeah, I’m a busy guy, but it’s all coming and isn’t that how the Internet works anyways?