|
[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]
On Wed, Nov 07, 2001 at 12:08:42PM -0500, Kyle R . Burton wrote:
| I've been reading up on is machine learning. One of the things I've been
| toying with is the ability to generate a regex to match a given example
| set of data. My particualr examples would be for things like phone numbers,
| or zip codes, or information that consists of single data elements.
this code looks very interesting. here's a thought off the
top of my head, dating from 1999 when I was doing coursework
in a related field.
i wonder if it would be feasible to evolve your regular
expressions as genetic algorithms in the usual hill-climbing
way. instead of refining a single pattern, try keeping a
stable of, say, a hundred possible patterns that each may
match only a certain subset of the input data. refine each
pattern through random mutations, so that a pattern from a
parent generation produces more than one child. the
selection pressure would be based primary on success against
the input data space and secondarily on length of regexp.
**Majordomo list services provided by PANIX <URL:http://www.panix.com>**
**To Unsubscribe, send "unsubscribe phl" to majordomo@lists.pm.org**
|
|