« Ind. Decisions - "Attorney fees, treble damages to be sought by state in sidewalk case" | Main | Ind. Decisions - "Ex-lawyer gets 2 years in deadly DUI," may serve 11 months »
Wednesday, June 03, 2009
Law - What exactly is "data mining"?
The Fort Wayne Journal Gazette today reprints a useful article on data mining first published in the Hartford Courant. Brief quotes:
But our growing digital footprint is threatening our ability to dodge inappropriate inquiries. Through data mining, employers, insurers, advertisers and others can infer the answers to private questions without even asking.They need two things: a heap of personal data and the techniques to crunch it. Both are readily available. * * *
Data mining relies on the principle that certain information – though useless in isolation – can take on new meaning when viewed en masse, or combined with other data. Scientists already use this technique.
There are two main approaches. First, data integration involves combining different types of data to learn something new. Consider a photograph of a bicycle: Alone, it’s an abstract representation. But tag the photo with your home location and a time stamp – and a public listing identifying the bike as stolen – and suddenly it becomes very meaningful.
A second approach is data aggregation. Gather enough of a certain type of data, and trends emerge. For instance, a cell phone’s location can be determined by tracking its signal. By aggregating enough location data from a single cell phone, we derive an increasingly reliable map of one person’s regular routes of travel. From this, we can estimate where the phone’s owner is likely to be at a given time and perhaps even guess his home location, income and so forth.
Fusing these approaches is even more powerful: that is, combining and mining multiple data sets, each very large. Google did this last year, pairing aggregate Web search queries with location and timing data to predict which regions would next come down with the flu. It outperformed the Centers for Disease Control and Prevention.
Posted by Marcia Oddi on June 3, 2009 01:34 PM
Posted to General Law Related