Klarinet Archive - Posting 000197.txt from 2006/05

From: Joseph Wakeling <joseph.wakeling@-----.net>
Subj: Re: [kl] Fingering and tuning charts, again
Date: Sat, 27 May 2006 20:45:51 -0400

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Tony Pay wrote:
> In view of the subsequent discussion, I found it interesting that:
>
> users.adelphia.net/~bmcgar/
>
> ...shows up as a live link in gmail.
>
> I wonder what triggers that?

Probably that Google has invested time and effort into creating better
semantic parsers to identify links in plain text. After all, their
search engine works on the principle of identifying how many incoming
links a given website has. A major benefit of gmail from this point of
view must be that they now have a huge database of *the links that
people are sending each other*, so it's in their interest to be able to
effectively identify as many of them as possible. And once identified,
give the user the benefit too ...

(The further benefit of gmail, from Google's point of view, is in the
textual content of emails. By allowing you to keep a near infinite
quantity of mail, they have a huge mine of semantic and sociological
information to analyse and base services on...)

Anyway, here's a little test to see if we can identify some of the way
their URL parser works:

http://users.adelphia.net
users.adelphia.net
http://users.adelphia.newt
users.adelphia.newt
users.adelphia.newt/~bmcgar/

The idea here is to test (a) whether having a valid high-level domain
(.net, .com, .uk, .dk, .de ...) in the text helps determine whether
something is considered to be a URL or not, and (b) whether having extra
bits like http:// or /~bmcgar/ added on helps tip the balance. :-)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org

iD8DBQFEePJUcjylL0sfzuERAqFaAJ9eoryRZOmVs/mg/0jiFYQI8Eo3eACeM4iD
NbSZBF9lSuGXOF/g+9jevv4=
=byLM
-----END PGP SIGNATURE-----

-------------------------------------------------------------------
Klarinet is a service of Woodwind.Org, Inc. http://www.woodwind.org