Friday, 13 April 2007

Google Alerts

Regular readers, if there is such a person, will recall my experiment before Easter with Google Alerts.

For those interested in how my Google alerts have turned out, it has been a mixed experience. On one hand, one Google alert did turn up several references to this blog, so clearly Google is indexing it. On the other hand, it also turned up a really random selection of references, some of them going back a number of years - a 12-year-old conference paper, a radio interview, and a short paper on The Fabulous LILO, Swinburne's metadata application profile. Why?

Try it out for yourself here.

1 comment:

Dana said...

The issue here of course, is that computers have not yet passed the Turing test. I think the alerts are based on Google's web crawlers finding new things, where the definition for "new" is "new to the index", so if a site moves, or the crawler bumps up against it for the first time (the web is a big place after all), then to the crawler it is new; the crawler isn't smart enough to read the text and find out what the real truth is (just like the crawler isn't smart enough to distinguish between Swinburne the institution and Swinburne the founder).

I personally have some Google news alerts set up, and most of the stuff I get off those is actually news, presumably because this space is better defined, more visible, and more comprehensively crawled than "the web" Occasionally I get a weird thing that is five years old and has just been moved into an archive, but usually what I get is an interesting read.