Corpora: statistics in CL question

From: Alexander S. Yeh (asy@mitre.org)
Date: Tue Mar 28 2000 - 01:49:54 MET DST

  • Next message: Ted Briscoe: "Corpora: Lectureship in Computational Linguistics / NLP"

    Recently, I saw the following statement (author is unknown):

    >In most studies of z-scores and t-scores in computational linguistics,
    >you tend to find that scores are too high. When you compute scores
    >for bigrams, for example, you would expect 5% of the scores would be
    >greater than 1.65, but you tend to find more than that.

    I am trying to find the studies referred to, and what makes some people
    believe that the scores are too high. Thank you.

    -Alex Yeh (asy@mitre.org)

    This archive was generated by hypermail 2b29 : Tue Mar 28 2000 - 09:13:29 MET DST