Getting started

So I am interested in helping make the Semantic Web vision a reality. Or at least doing research on interesting problems related to the Semantic Web vision.

This means I am interested in ontologies and formal knowledge management and engineering. This is my “bread and butter”. Then along come all these people attacking the very heart of that whole business with the very strong claim is that formal management is dead and is replaced by “organic”, “evolving”, informal user centered “management”.

Ontologies vs. Folksonomies. Well, that is the strong description … but not an unfair one if you take the extreme view: Clay Shirky. There are, of course replies to this. There are too many links, but a good starting point is here.

So are tags better than categories? Is it better to tag or to categorize? The argument is heated, for sure. BUT … one thing I have not seen settled is .. the most fundamental question that MUST be asked even before such a debate can take place! What is the relationship between tagging and categorization? Are they really two different activities, such that we can put them on opposing sides of a battle? Certainly people have strong opinions on this … opinions usually supported by single examples or plain rhetoric. For example in the Shirky blog cited above, he shows an example of the tag “to_read” and points out that such a tag would never be used in a formal taxonomy. But it is a perfectly good tag, it is claimed, because it means something to the user.

Well. Does it really? How many things can you label “to_read” before your bookmarks start to look as messy and useless as the desk in my office?? At least there, the (must read) magazines and articles from last year are hidden by the accumulated mess on top. But not on a flat web page!!
I have seen no studies about the effectiveness of tags, in general, in retrieval or even browsing. Is there even a useful way to measure whether or not tags are useful? (Apart of course from the number of sites using them and generating advertising revenue!) Are tags like “to_read” useful? And besides, how common are tags like that? Is THAT the only example, or is every tag pretty much like that. Which reminds me … just what ARE tags like??
These questions have not been asked so much. Have they? Why not? We have many observations about statistical distributions, historical trends and mathematical analyses. But nothing in depth about what these things are, and how they manage to come out of people. This is when I start to get excited 🙂