If a user doesn’t care about the boundaries, they should be
transparent
Use of LCSH or own subject categories, controlling vocabulary
Users getting finding aids instead of content, having mix of text and
images
One researcher made the comment that …
I have a nice example from the PA Gaz, an 18thC newspaper that was the NY times of its
day
I see this traversing the yahoo
directory tree, where some directories have one subdirectory with 10 items
next to another subdirectory with 10,000 items
Idiosyncratic: under Soc, Manners & Cust we see the subtopic Venom?
say latent semantic indexing and latent semantic analysis same thing
say probabilistic clustering (e.g. pLSA)
say latent semantic indexing and latent semantic analysis same thing
say probabilistic clustering (e.g. pLSA)
say latent semantic indexing and latent semantic analysis same thing
say probabilistic clustering (e.g. pLSA)
say latent semantic indexing and latent semantic analysis same thing
say probabilistic clustering (e.g. pLSA)
MetaCombine used a scheme similar to LSA to do their topic clusters
The topic model was developed specifically for text problems