A client recently forwarded a blog post to me about folksonomies and asked if this is something we should consider.
Here is my take: Social Tagging (use of “Folksonomies”) has a valid place in the scheme of things. We can use them as a source for candidate terms or new term harvesting. They are also useful for content that is less structured that may be tougher to organize (discussion or blog postings for example) or material that does not justify structured tagging. (Obscure web pages). The fundamental issue here is that they don’t take the place of formal taxonomies but can be used to augment them. In some cases, user generated tags make a lot of sense. (A group of engineers working on a new product might come up with terms that are not yet in the formal taxonomy. They also speak the same technical language and use the same terms. Raytheon uses this approach of social tagging for what are called “featured results”. These appear along with the “officially” tagged content).
Here are the advantages to a folksonomy (or social tagging) approach: Adaptability – new terms can evolve quickly and be applied to new concepts
Lower cost – many hands make light work – by distributing the workload amongst a large number of people, there is not a lot of burden on a central group
Flexible – anyone can tag anything with anything so there are no rigid constructs
Takes into account multiple perspectives – you and I may use different words to describe the same thing. If we both tag according to our understanding, both of our points of view are taken into consideration
There are a number of disadvantages to this approach:
Over tagging – too many tags on content can make search and retrieval meaningless – recall is high, but precision is low (lots of irrelevant results)
Inconsistency – Misspellings, different punctuations, capitalization, variations in spelling all show up as different tags and cannot be used for filtering and navigation
Skewed perspective – Few people tend to actually apply tags and they can be over represented
Lack of control – for the last couple of years if you searched Google for “Miserable failure” George W Bush’s bio was the number one result. Though the mechanism was somewhat different, this was still an example of user generated tagging. With social tagging, users could conceivable tag company content with negative terms A corporate incarnation of this would not have the same risks. We would control the terms to some degree. (A taxonomy by its nature is a “controlled vocabulary”. Folksonomies are “uncontrolled vocabularies”) The hybrid method (freedom of social tagging with some control and vetting processes) is actually something we are planning on in our taxonomy roadmap. Basically we want to take in new terms that users would suggest and then put these terms through a review process. We would make it very easy for people to suggest new terms and then put them into our taxonomy management process.
Some people might argue that this is an example of typical taxonomy development, but the different would be the sources of tags and the ways in which tags would be applied. A feedback mechanism would keep terms fresh and nominate new terms while “voting” on the appropriateness of existing terms. Some of this would need to be integrated with the content management system, but some of this functionality is in Wordmap (the taxonomy management tool). Not a lot of people have done this yet, so it is very leading edge. There is also something called “social search” that allows user behaviors to influence search results. I am still looking into that and can provide an update when I learn more.