Re: Corpora: Negative mutual information?

From: Miles Osborne (osborne@cogsci.ed.ac.uk)
Date: Thu Mar 08 2001 - 14:39:46 MET

  • Next message: Ted Pedersen: "Re: Corpora: Negative mutual information?"

    a negative mi sounds like a bug. as far as i understand, mi can be
    thought of as the kullback-liebler divergence between an independent
    and a dependent model. kl diverengence is always postive.

    (also check Manning + Schutze p 72)

    Miles



    This archive was generated by hypermail 2b29 : Fri Mar 09 2001 - 01:15:55 MET