#1 WordProbability refactor

closed
nobody
None
5
2003-07-16
2003-07-09
No

WordProbability can now consist of either:

1. A Word with a set Probability or
2. A Word with with matching & nonMatching counts -
the probability is calculated by this object.

Not happy with the "normaliseSignificance" method
location. Any suggestions?

Discussion

  • Peter Leschev

    Peter Leschev - 2003-07-09

    Logged In: YES
    user_id=752383

    Forgot to mention, this change is dependant on commons-lang
    - Hope you don't mind!

     
  • Nick Lothian

    Nick Lothian - 2003-07-09

    Logged In: YES
    user_id=763364

    I need this in unified format (I think!). I'll apply ASAP.

     
  • Peter Leschev

    Peter Leschev - 2003-07-11
    • milestone: --> 297425
     
  • Peter Leschev

    Peter Leschev - 2003-07-11

    Version 2 - Has more refactoring than just WordProb.

     
  • Peter Leschev

    Peter Leschev - 2003-07-11

    Version 2 - New files associated with refactor.

     
  • Peter Leschev

    Peter Leschev - 2003-07-11
    • milestone: 297425 -->
     
  • Peter Leschev

    Peter Leschev - 2003-07-11

    Logged In: YES
    user_id=752383

    Ok, here's version 2 in the unified formatted diff.

    I've continued my refactor:
    - I've created an ITokenizer interface and pulled out the
    tokenizer code out of the BayesianClassifier as other
    classifiers would need to use tokenizers as well.
    - I've created some unit tests cases as well...

     
  • Peter Leschev

    Peter Leschev - 2003-07-16

    Logged In: YES
    user_id=752383

    This patch has been committed to cvs.

     
  • Peter Leschev

    Peter Leschev - 2003-07-16
    • status: open --> closed
     

Log in to post a comment.