#47:	Query-directed Data Mining using Python and Parallel Processing
Presented:	Thu Mar 24, 2005	Sprinting Room 307	03:30 pm-04:00 pm
Author(s):	Christopher Gillett / Compete, Inc.
Items:	audio-no    handouts-yes    released-yes    video-no		ADMIN
Abstract:	Compete, Inc. has built a query processing system using Python which forms the basis for all query-directed data mining within our company. The system accepts queries written in "near SQL" syntax and performs parsing and analysis of the queries. Query decomposition allows exploitation of parallel computing resources and allows queries involving large amounts of data to be answered comparatively quickly. The system was built entirely using Python and is used in conjunction with the Portable Batch System and various Unix utilities.
Item(s):	QueryDirectedDataMining.pdf 20:19:23 2005/04/05 46.5 kB application/pdf abstract.stx 02:58:34 2005/04/14 537 bytes text/html

Note: Talk recordings have come from different donors, with different levels of quality. A suffix has been added to the basename of each recording reflecting this. For eventual upload to a repository like archive.org, a formal naming convention has been followed:

pycon-{date}-{track}-{timeslot}-{talkno}-{donor}.mp3

For those who might prefer a more human-meaningful name, the recordings have MP3/Ogg/Flac ID3 information within and a simple python script could rename your collection to something in a {title}-{author} form.