Talk:Data mining

From SourceWatch
Jump to navigation Jump to search

These definitions of "data mining" don't look any different than standard "data analysis" which is the kind of things databases were created to enable. For example, "show me all the orders placed for widget-A, sorted by time" can result in "Look! orders are steadily decreasing". Standard analysis activity.

Now, to me anyway, in order to have a new activity, not just a new name for the same old thing, data "mining" would entail, um.... "mining"... of data by extracting selected fields from several different, unrelated, databases to assemble new database records. For instance, the mining of SSN/name association from one database, and the name|or|SSN/drivers_license_number from another database, and the name/phone_number from another database, and the phone_number/purchasing_record from another database, and the SSN/credit_card association from another database, etc. and *primarily* (if not exclusively) for the purpose of compiling information records on unexpecting and unwitting individual citizens.

It could also be used however to associate illness with pollution with chemicals with manufacturers and other civic minded projects.

Note that a set of relational databases can be maintained, or integrated, such as by the MIPT Terrorism Knowledge Base without "mining", but just by standard relational analysis.

Additionally, "mining" can entail the addition of new data fields or records or bases by compiling written notes from disparate sources such as medical or pharmaceutical records.

While we're nitpicking ... data mining and data integration are two separate processes ... one searches and sorts, the other, well integrates. To put this very simply, it is the difference between sorting out the red jelly beans as opposed to integrating the red jelly beans in with any number of other-colored jelly beans.

The former extracts information from a data set and the latter integrates various data bases into a master file. Not at all the same thing.

Now, analysis, is another thing altogether ... that's where the nitpickers come in who are not only looking for a certain type of red jelly bean in the bag but also want to know where it came from, who it knows, what shade of red it is and how it got that way, etc. Artificial Intelligence 16:57, 16 Jun 2005 (EDT)