Rants and musings of Paul Scott
Paul Scott on 2010-10-14 18:32:39
Moving from folksonomy to ontologyThis is simply a short summary of a few ideas around folksonomies that I had recently.
OK, so folksonomies are pretty rife and easy to come by, but the main problem with coming up with usable data from human generated tags is, well, the human bit. So what do I mean? Well, humans are human and are inconsistant by nature. This is one trait that makes us human right. Ontologies and structured data is structured and formalised. This leaves a pretty big disconnect between the leaves of popularity and the tree of knowledge. There is, however, an evolutionary process that will take place in longer standing folksonomies that make them start looking like ontologies. Lets look at an example rather.
We will start with the FIFA World Cup held in South Africa this year. Before the tournament started, there were some debates as to the correct tags to use as the official tag. There were a few, that all described the event quite well, but were inconsistant with each other. Examples include WC2010, fifawc2010, WC10, SWC2010, SWC, worldcup etc. All looks good, but essentially useless when creating structured data from the information contributed by the people using the tags. When FIFA decided eventually on an official tag, most people then migrated to use that tag in order to be aggregated to various social aggregation systems and to communicate their message more effectively. This evolution from highly unstructured data entry to a more structured consistant format is what I am talking about here.
The above example is in no way unique. Just take a quick look at the "Trending Topics" on twitter to see this evolution in action. The trending topics have generally evolved from the general noise to somewhat structured data entry points, much in the same way that a swarm of insects, fish or herd animals will organise into a structure of sorts by nature in order to offer protection etc. The semi structured entry points can then be somewhat easily extrapolated to a more ordered version with vocabulary and threading.
This can be demonstrated quite easily with some simple code and a "Game of life" style simulation. This will almost always happen.
Businesses and market analysts should be looking at the way that things evolve and predicting where things are moving towards, rather than trying to shove tired stuff down the throat of a reasonably savvy and individualistic crowd. The individuals will indeed find common ground and group up eventually, so look for the signs that they are about to and profit!
B. F. Skinner and the theories of radical behaviourism definitely apply in the social network. There will always be an "in group" - i.e. the trending topics, and an "out group" - those that do not (yet) participate in the current trends. The out group may tend towards the in group, but with such a large population, there are multiple in groups, which adds a little complexity.
