Here’s a question for any experts on learning object repositories, metadata, cataloguing, etc. Given the latest conversations around tagging, folksonomies and Google’s various search tools; is it still necessary to create a definite metadata structure for large web databases? Would it be better to focus on search? Can personal tagging address everyone’s needs? Should you address both? What would you do if you had to build a learning portal tomorrow? Yes, I’m asking for free advice ;-)
Update: Stephen Downes has this Summary of the Metadata Enigma, with the final comment being that the benefits of metadata have not been clearly established.
maybe, not quite yet IMHOHere’s one way I illustrate this issue for people. There is a photo of Alan Levine and my kids in my backyard from when Alan and his wife visited us a few years back. It is posted in my flickr site. Find it. Have trouble finding it? Why? (There are a few reasons I think). Much of what ‘folksonomies’ and other approaches seem to me to have enabled are wonderful serendipities. Which is great and useful – it is always exciting to be on the general hunt for something and then to have a resource that would fit the bill fall in your lap. This is a different scenario, though, to ‘I want to find this thing and this thing only’ which I think is what I am describing and for which very loose metadata, very loose cateloguers and the like may not get you the result you hope for. Both of these use cases are important and one shouldn’t be ignored over the other. My other main complain with some of these approaches is that they depend on the resource existing on the "open web." Think of google. It’s not to say that google doesn’t have its own algorithims for parsing the relevance of a document solely on what is contained in the document, but its power comes from looking at it in the context web pages that link to it. It seems to me that when you start talking about resources that don’t live out there on the open web, then this becomes more problematic, and like it or not, password protected material is going to be with us for a while. The flip side isn’t that ‘folksonomies’ and other contemporary approaches aren’t any good, far from it. It just feels to me like there is still a role for now for metadata, cataloguing, etc., and that for now the two need to coexist. I guess in part the answer to your question depends on your time horizon. Are you asking philosophically, or do you need to implement something tomorrow?. Given the rate of innovation we are seeing in this and other arenas, it seems foolish to make predicitions that are very far off in the future, and there are advances on things like automatic categorization, keyword generation, semantic anaylsis, etc, happening every day. So it may be that a combination of increasingly powerful search technologies and lightweight tagging practices are all we end up needing. For now though, I don’t think I can fulfill my own uses cases in the repository world by just relying on that. Just my $0.02 worth (which the way the Canadian dollar is going is almost worth $0.03 now!). Cheers, Scott
Infolibrarian Corporation is proof that metadata is not dead. Companies are not using the right tools to achieve success in their efforts to capture and catalogue their islands of information. Infolibrarian Corporation have successfully implemented hundreds of metadata programs using Infolibrarian software and hardware. In our space, metadata is alive and well…
Keith Brewer -CEO
Infolibrarian Corporation
160 Linden Oakes Park
Rochester N.Y. 14625