Re: Ternary search trees for Unicode dictionaries

From: Doug Ewell (
Date: Thu Nov 20 2003 - 11:12:52 EST

  • Next message: Philippe Verdy: "Solving a regular expression ambiguity for Hangul syllables"

    Philippe Verdy <verdy underscore p at wanadoo dot fr> wrote:

    > The question of processing performance may not be very relevant to
    > create a lexical dictionnary, where what is relevant is the final size
    > of the dictionnary file. As long as the compressed data can be
    > accessed fast, the size of the intermediate tables and the time needed
    > to compute the dictionnary structure will not be decisive.

    You are correct. I was looking at the bigger picture of data
    compression rather than your point about creating a data dictionary on a
    one-off basis. It is certainly true that if you are creating some sort
    of data file once, to be used over and over by end users, the time it
    takes to create the file is a relatively minor consideration.

    As a matter of fact, in my upcoming paper I basically ignored speed
    considerations, except to mention that they exist, because they are
    often difficult to quantify in the abstract. It's easy to make a
    blanket claim that Algorithm A is faster or slower than Algorithm B, but
    to prove this one must compare specific implementations, single- or
    multiple-thread, running on specific operating systems, on single- or
    multi-processor hardware, etc.

    -Doug Ewell
     Fullerton, California

    This archive was generated by hypermail 2.1.5 : Thu Nov 20 2003 - 12:12:38 EST