Greg Detre
Tuesday, December 10, 2002
NL
information servers
medstract.org
they�re
looking at the domain of bioinformatics
enormous corpus � 10m papers, 40,000/month, 36GB
also clinical databases, microarray databases, proteanic databases
building an
ontology of entities, events/processes etc.
they�re
trying to automaticise the process of adding structure at lower levels than a
given leaf
siblings on the tree are not necessarily exclusive � this is part of the
type mismatch problem � they could just be different views of the same thing
MB learning???
hyperbolic visualisation??? support vector graphics???
generics???