Rod Boothby triggered these thoughts in his presentation (available as a webcast) at the Collaborative Technologies Conference. He presented the concept that the as the number of people who tag increases above 50 you quickly cease producing unique tags. This is backed up by research from HP. I find this concept and the research from HP exciting and as Graham has also blogged there will be some interesting work to make this a standard in the enterprise. The faster we can encourage people to tag all their work, and the quicker we can encourage vendors to have open document formats that support tagging, then the faster that we will start to see the power of tagging, and more importantly the folksonomy that produces in the enterprise. There are a number of papers on social bookmarking producing folksonomies, this paper examines the IBM Dogear system.
The folksonomy clouds that we presently observe are generally number oriented e.g. I have tagged six documents with this tag. In the future I would find it hard in an enterprise just to use a simple counting mechanism. Time must come into the folksonomy and as the period of time passes the relevance of that document, and hence its weighting within the cloud, must decline. The whole concept for the enterprise is to find people, teams, documents and projects with similar themes that are happening now or have recently completed.
So the challenge for people like myself who influence both vendors into ensuring that users can tag with their products and influence enterprises that this new method of social bookmarking makes folksonomy a must for their enterprise. The trouble will come in the fact that many enterprises have categorised discussions and they simply exploded into a degenerating mess of categories, now what we need to do is say to them that the mess is fine because once you use the categorisation to produce s folksonomy cloud it becomes much easier to navigate and much more managable – especially if the systems which vendors develop will identify and suggest tags from within the content.
As an aside I have now updated the sidebar of my website to include tag clouds of what I blog (through a wordpress plugin) and what I read (through the tag cloud facility from del.icio.us). And a final note to let you know that the quiet blog was due to holidays nothing more serious as one reader thought!
Hi Stu,
Think the link to del.icio.us is incorrect – goes to IBM.
cheers
Charlie
LikeLike
Well spotted that man! fixed!
LikeLike
Have you come across a good time-based tag visualization yet? Yahoo demoed an interesting one at CHI2006…
LikeLike
Sacha, I hadn’t seen that visualisation tool. I will look into that thanks.
LikeLike
Hi Stu,
the time dependence of tag clouds is a very important issue with current tag clouds and it is deeply related to tagging decay and emergent tags.
More details on them on my blog
http://www.infospaces.it/wordpress/topics/general-stuff/82 and on Bokardo (http://bokardo.com/archives/web2con-popularity-decay-in-tagging/ and http://bokardo.com/archives/web2con-emergent-tags/)
By the way, imho the overall concept is that having a flat tagging space, sorted only by an arbitrary criterion (alphabetically with counting), is not enough to provide scalability and to make tag clouds useful tools for end-users.
More on my blog and on the FaceTag project (www.facetag.org).
Cheers,
Emanuele
LikeLike