← Back to team overview

wintermute-devel team mailing list archive

[info] (part 2 of 3) Wintermute's Data Mining

 

Data mining is perhaps Wintermute's biggest tool. Without its ability to 
obtain information about its environment, Wintermute only knows what it 
does within the COSMO ontology. And that, compared to the vast amounts of 
information out there in the world, is NOTHING. Another beautiful aspect of 
Wintermute, inspired by KDE, is modularity. We can define plug-ins that ease 
the process of data mining for other developers and provide connectors 
(very similar to how Zeitgeist for the GNOME desktop had plug-ins for 
xChat, Banshee, Synapse and a collection of other applications) to 
currently existing programs. Of course, this approach would only expose a 
limited amount of information (last used files, currently open windows, 
etc). 

We would need the following sorts of plug-ins to be implemented for data 
mining to provide a basis for data mining:

	- BaseDataMiner: This miner would provide caching abilities for all
	  mines. Any capability that we think all mines should have should be
	  exposed in this plug-in.

	- FileDataMiner: This miner provides a means of obtaining meta info-
	   rmation about files. This should work similarly to how the Message
	   class in WntrNtwk can contain information from its derived forms so
	   that binary and file data would be compared carefully. Note that this
	   plug-in may need Web capabilities for file uploading, downloading, and
	   querying.

	- WebDataMiner: This miner eases any other miners that want to obtain
	   information (like web pages) from over the Web. 

Generic miners like "GenericWeatherMiner", or "GenericCalendarMiner" could 
provide a set-up for calendars and weather information obtaining sources. 
This still needs a tad bit more thinking and drawing up.

	   
-- 
Jacky Alcine <http://www.jackyalcine.co.cc>

Attachment: signature.asc
Description: This is a digitally signed message part.


Follow ups