(116) previous ~ index ~ next

---

To: Rich Schwartz <schwartz@bbn.com>
From: Jaime Carbonell <jgc@NL.CS.CMU.EDU>
Subject: Re: Training data for dry run?
Date: Fri, 11 Jun 99 16:49:31 EDT

---

Jon, Rich,
Clearly you don't need English background models (TDT1 or TDT2 should do
to construct these on similar corpora).

Best would be Rich's suggestion of other (unlabeled) Mandarin text,
however produced. It's not good to "waste" part of the corpus, especially
the part that may have the most topics represented.

--Jaime

---

(116) previous ~ index ~ next

Last updated Mon Jun 21 11:18:49 1999