Tuesday, August 14, 2007

Some progress with and some motivation for ED

Eddie and I have made some good progress on the ED application. We've got the first set of bugs taken care of and are getting ready for the first big public announcement. The only remaining must-do is the completion of the query expansion script. When that is done and tested (at least twice!) I'm going to move the database / servlets over to a dedicated computer and spam my friends..

Lots of people, including my supervisor, have expressed some uncertainty about why I am doing this - so I think it might be a good idea to try to clarify my motivations a bit. Perhaps this will even help encourage people to try it out.

The central thrusts of my research are to 1) identify ways that 'ordinary' people (non-computer scientists) can help to build the semantic web and 2) to motivate them to do so. One of the fundamental pieces of the semantic web is the description of web resources (URIs - especially URLs) with concepts from web-accessible ontologies. Thus, it would be great if I could figure out a way to get normal people to start forming these connections.

In social tagging applications, like Connotea and Del.icio.us, I see a tremendous number of people contributing a new, significant, but inherently limited form of meta-data to the web in the form of tags. The motivation question does not seem to be a problem here. They are putting their descriptions of the web (metadata) online, for free, but the problem is that these descriptions are represented in their own individual, unstructured, unshared, typically unplanned and unconnected terminologies (tag collections) rather than the shared, structured ontologies that might someday form the backbone of the semantic web. ED is an attempt to build a new variety of social tagging application that continues to satisfy the personal information management needs of the populace while at the same time producing information collections that are more useful, both to the individual taggers and to the semantic web community through the use of ontologies.

Would love to hear what you think about all of this!


Pedro Beltrao said...

This might be naive but I maybe once a group of people start using this it could be possible to start annotating some of the folksonomy tags with terms from structured ontologies. If the tag X is always present with GO term Y then maybe they are the same and you can link the two.

Benjamin Good said...

Hi Pedro, I don't think its naive at all. Its similar to feedback I reveived from Robert Stevens who is definitely not naive. He immediatley envisioned an extension that would allow the users to start building relations - particularly hierarchical relations between the tags right in the same context as when they add tags. I think that there is a lot of opportunity here to start building or at least contributing really useful feedback to creators of terminologies. Even without ED, it would be possible to start allowing users to build associations between tags - like synonomy as you suggest. Bibsonomy already lets users add narrower-than relations between tags and I don't think there is any reason to stop there.
The other notion that I think you are suggesting is to start adding associations like this automatically. I think a combination of automatically generated suggestions with user-moderation might work very well.
The to-do list for this project seems to be growing by the hour ;)