I was slightly on the wrong track in a previous blog, I think. It happens …
So, the mistake was that I was thinking that all of the popular tags that had survived in a historical context, could be thought of as “category labels” and the less popular ones were the highly individual ones. This would certainly have made things easy. Of course it is good that this is not the case, because that would have been TOO easy. And things that are easy, are not interesting. So why is it not true? Just look at the table below, which shows the top four most popular tags for some of the 50 most popular sites on delicious:
| Site | Slashdot | Flickr | Pandora | Digg | BBC News | New York Times |
| Tags | News, Technology, Geek, Daily | Photos, Flickr, Photography, Sharing | Music, Radio, Recommendation, MP3 | News, Technology, Blog, Daily | News, BBC, UK, Daily | News, Newspaper, Daily, NYC |
It is self evident that the popular tags form a heterogeneous collection. Some are clear category labels that would feel at home in a formal taxonomy (e.g. “News”, “Movies”, “Music”). Some, like “Daily” and “Recommendation” appear to describe resources with a particular property which is nevertheless fixed and user independent. Others like “Fun” and “Geek” describe more personal properties that depend on individual interpretation. Finally there are proper names like “UK” and “NYC”.
So, what are the facts telling us???
technorati tags: ontology, folksonomy
Csaba's Blog 
How people tag content is only half the battle. How to people later use the tags to find content? Do they really use tags or categories, or do they primarily rely on search?
Any idea what percentage of users use search vs. drilling down through tags or categories to find things?
Excellent question, Ed. This is certainly something I have been thinking about, but I have not come across any direct evidence that bears on the question. I have seen some experiments that deal with related questions, but nothing direct.
As far as I am aware, the kind of information we would need to address the question can’t be obtained from delicious. If this is not true, someone, let me know!!
In the meantime I have emailed delicious support about getting more data, and apparently they are working on a “research license” to allow more access to data. Who knows how long that will take? Maybe more people should write??
In the meantime I suppose we would have to run an experiment on one of the open source versions of a delicious-like application…
There is a paper written by Kristina Lerman and Laurie Jones called “Social Browsing on Flickr” that suggests that “social browsing through the contacts’ photo streams is one of the primary methods by which users find new images on Flickr”. The paper is very interesting though I think that this is a specific characteristic of Flickr and may be totally different from del.icio.us.
You can find the paper here: http://arxiv.org/abs/cs.HC/0612047