Tuesday, March 8, 2005

Disappoint-ensity

When a watchful mentor emailed me a link to Attensity over the weekend, I was encouraged, finding, from a quick glance at their web site, that some of the same data-mining they market, as their hallmark, matches up with a few of the configurations defining a project I have underway.  Attensity claims to process data and analyze that which is otherwise difficult to discern.  Their software churns away like some kind of high-powered heuristic meaning-cruncher--a processor of mass quantities of text into readable metadata.  NORA (non-obvious relationship awareness) for large-scale discourse. 

So I sent them an email inquiring about the whole plot, all-the-while-recalling that the only other text parser I looked up asked $2k per year for software licensing.  Did I mention I'm a grad student?

Well, I did mention that in the email to Attensity, the email inquiring about their project.  And I heard back today--a polite note, something about serving the US Intelligence community and a starting fee of $50,000. And something about wishing me the best in my quest.  Thanks.  But no thank-you. 

The proliferation of textual analysis apps self-identifying as the devices built to root out terrorism piques my interest.  Perhaps because of the sheer volume of text to be analyzed for particular patterns (suspicious patterns! watch the parentheticals, Attensity!), mass-discourse text parsers are up and coming.  And so unbelievably over-priced that they're no use whatsoever to the project I'm working on.

Bookmark and Share Posted by at March 8, 2005 10:47 PM to Slouching Toward
Comments

Well, remember that the gummit (and not Al Gore) was responsible for the development of the Internet´┐Żat least that's the word around the campfire. So maybe this technology, too, will one day be used for some purpose other than to terrorize terrorists (she said, hopefully). Meanwhile, save your pesos.

Alternatively, all this might mean that your grad school experiences could lead you to truly lucrative employment (she said, with her tongue definitely in her cheek).

Posted by: senioritis at March 9, 2005 8:38 AM

Gummit? Gummite? Don't know that one. I'll continue to save my change but more likely for Ph.'s college days than my own software extravagancies. And lucrative employ is no trade-off for fulfilling employ, right? I'd have never left KC if I was trying to worry hard about money (just as I was contemplating selling blood to travel to SF for CCCC).

Posted by: Derek at March 9, 2005 7:19 PM

I thought it was the gummint, myself. But it sounds like there are two different issues you're looking at, Derek: one, the harvesting of the information, and two, the filtering of it. It strikes me that (1) is the hard part: after all, regarding (2), I've got colleagues who are putting together Perl translators for Old English, and Anne Herrington and Charlie Moran wrote recently in CE about computerized essay evaluators. If you've got an algorithm for making sense -- well, what are you looking for, exactly? Ever since reading Anne and Charlie's article, I've waited for the Slashdot post offering information about reverse-engineering the algorithm that "reads" and "evaluates" student essays and how to game it.

So I guess I'm asking: is it the volume that's your concern, or is it the patterns?

Anyway. Not quite sure how to read your last parenthetical -- I'm hoping that you will, in fact, be at CCCC, yes?

Posted by: Mike at March 9, 2005 11:04 PM

Yeah, I'll be at CCCC. I hope we have a chance to get together. I have your Friday morning session penciled in, but maybe we can meet up before then. I'm coming in on Wednesday mid-day and leaving out on Saturday mid-day. I'd like to hear more about the Perl scripts. I was on the verge of applying one to the current project, but I went with something simpler. I'm not turning it toward student writing, and I'm interested in both volume and pattern to some degree, but also a few other questions/devices/methods.

Posted by: Derek at March 9, 2005 11:15 PM