Folksonomy refers to public tagging or labelling and extracting/searching/obtaining data from that. You can take del.icio.us [Social Bookmarking site] for an instance wherin you bookmark sites under various categories and that is searchable by others too... To be precise, people sort contents into better categories than bots and that help others online to get more relevant data.
This became popular after these social networking and colloboration sites. And also it is true that people know what they want better that these automated bots do. So these tagging and labeling of various sites, webpages, photos etc helps others too... Folksonomy is the base on which Semantic Web is built, which has metadata instead of tags/labels in machine readable format.
There are many services built upon this theory.
Imagelabeler is a Collaborative Tagging Game from google where you label random images along with some unknown partner. So...if you and your partner tag the image with same name your point increases and another image is loaded. This helps to analyze what people exactly want. This in turn help their image search feed better results.
Flickr also enables users to search for images based on labels applied to them.
What about Meta Noise ???
That problem is yet to be addressed...
Links
http://www.folksonomy.org/
http://www.vanderwal.net/folksonomy.html
http://en.wikipedia.org/wiki/Collaborative_tagging
