The company I work for is building a gi-normous database of everything you ever wanted to know about any product, company, or ingredient you've ever heard of. As it turns out, getting lots of data isn't hard, but combining it all together is pretty hard, and making sense of it is very hard. There's a lot of data in the world, no perfect universal product identifier (UPC isn't specific enough for our purposes, and most data sets we get don't include it), and the data is very dirty. In the process of building smarter data ingestion systems and massaging the data, we often come across some hilarious bits of it.
Some things that have turned up on ingredient lists (aside from the usual cryptic 12 syllable chemicals):
- dust mite
- panther fur (in a bath soap, of course)
- air
- tobacco dust
- bone oil
- thistle
- DNA
- "explosives" (no further elaboration)
- care (but, surprisingly, not love)
There are also some hilarious mispellings, often of "deionized." My favorite was "demonized water." You gotta watch out for that stuff.
And there are also the hazard lists. Hilarity ensues when bored government workers have to make official lists. Workplace hazards include "shift work that involves circadian disruption" and "adding machines."
1 comment:
Hi Ann, are you using some particular data mining algorithms ? If you need to extract from your huge DB significant data, you might try with WEKA3 (data mining field). Inside you'll find lots of "well-known" data mining algorithms ready to be used !!!
Lemme know if you need more information !
Post a Comment