Folksonomies

The term folksonomy refers to a decentralized approach to classification or classification by the folk if you will. Read this paper for some details. Given the amount of data generated on the web each day one would need an extremely fast and scalable technique to classify data. Automatic classification a la data mining/machine learning techniques are deployed more for small scale structures, if at all. A manual decentralized approach is what is popular for the internet today. Actually, I should restrict myself to the social networking parts of the internet. Take the bookmarking website del.icio.us for example. Each user has a list of bookmarks. S/he tags each bookmark with some keywords describing them. The user can now retreive links based on keyword searches. Not only that, this simple feature of adding metadata allows the various links listed on del.icio.us by various users to be classified, ranked etc. This can be used to study various aspects of the social network formed by the users of the website. As a result, not only is the data being classified, but so is the user. And this can be a primary tool for a web services company seeking to provide a personalized experience. A customer at Amazon.com for example, who also has an account at 43things (a company with amazon as the only(?) investor), can get a more personalized service based on his goals listed at 43things. The same applies to Yahoo! and Flickr and possibly Google and Blogger. Here is a paranoid but funny view of how things will end up if user generated metadata continues to increase.