Global term vector exists?

[Fredrik Andersson <fi...@gmail.com>]

Hi gang!

Is there an accessible global term vector of all encountered terms in a 
Lucene/Nutch index, or d'you have to build this yourself by enumerating all 
the documents and gather their individual terms? I'm also wondering, has 
there been any previous attempts to make Nutch use a latent semantic 
indexing approach (matching queries by vector angles rather than keywords)? 
The map-reduce framework could really come in handy in this area.

Greetings,
Fidde