Revision history for Perl extension Lingua::Wordnet.

0.01  Mon May 17 10:58:40 1999
    - original version; created by h2xs 1.19

0.1   Mon Feb 2 1:28:30 2000
    - rewrote whole module to convert dbs to Berkeley DB files
    - took out all Wordnet C API code

0.2   Fri Apr 14 11:24:30 2000
    - added pointer counting to 'scripts/report.pl'
    - fixed examples for Lingua::Wordnet::Analysis::intersection() to 
      use references
    - added Lingua::Wordnet::Analysis::distance()
    - changed Wordnet.pm and scripts/convertdb.pl to allow custom data
      delimiters via the $DELIM and $SUBDELIM globals.
    - made nested subs in Analysis.pm anonymous, thanks to Josef Meyer
    - fixed lookup_synsets morphing, thanks to Melissa L. Holcombe
    - changed morph() to require a POS, and convertdb.pl writes the 
      morph database with POS labels.

0.3   Sun Apr 16 3:24:12 2000
    - lots of POD edits
    - improved (fixed) lots of add/delete pointer functions
    - fixed Lingua::Wordnet::Analysis::match()

0.4   Thu Aug 24 11:36:14 2000
    - fixed installation to recommend a 'cd' to convert the dbs
    - removed plan to automate the synchronization of corrsponding 
      pointer types
    - more POD updates and typos
    - added a small 'wn.pl' script to 'scripts/' to illustrate the 
      most basic usage

0.5   Thu Aug 31 20:07:54 2000
    - added 'overview' method due to popular demand
    - overloaded stringify operators ("") for class 
      Lingua::Wordnet::Synset; you can print '$synset' directly 
      and get an overview for the synset; it uses 'overview()'
    - added test of overloaded behavior

0.6    Tue Oct 17 2000
    - Changed the overload syntax in Wordnet.pm (line 559) to work 
      for Perl < 5.6.
    - Fixed some typos in 'scripts/10questions.pl'

0.65   Wed Feb 21 2001

    - Fixed documentation errors
    - Added reverse_morph() lookups (don't use them)
    - Fixed errors in new_synset()

0.7    Wed Mar 7 2001

    * Data file formats have changed in 0.7. This requires new conversion using
      the 'scripts/convertdb.pl' script. What hasn't changed is the way any of 
      the indexes are written, so this should not affect most code. What has 
      changed is the way each word entry is written to the data files. Word 
      entries are now of the form "lemma%sense_num", where sense_num is the 
      unique number for the sense of lemma in a POS (rather than unique within 
      a lexicographer file). This makes the return values for words conform 
      to what you might expect to see when searching for a given sense.
  
    - Added lookup_synset_all()
    - Added scripts/big_struct.pl
    - Added scripts/tree.pl
    - Fixed the lemma% concatenation, which misrepresented the sense numbers
    - Changed 'scripts/convertdb.pl' to try harder to update paths in 
      Wordnet.pm

0.71   Mon Aug 13 2001

    * This release is compatible with Wordnet 1.7, but requires a rebuild of 
      the data files. Wordnet 1.7 changes include new file byte offsets; these 
      do not map to 1.6 offsets, and this module performs no conversion (no 
      access to the "Sense Map" package). Additionally, some sense numbers
      have changed.

      This release should still be compatible with 1.6, but includes no 
      features that enhance 1.6.

    - 1.7 does not have a "cousin.exc" list; caused the conversion script to 
      not choke as a result.
    - Added a script 'scripts/convertdb_win.pl' for Win32 compatibility; thanks 
      to Ray Schumacher.
    - Changed test.pl to work with new synset numbers

0.72	Mon Dec 17 2001

    - Fixed test.pl ... again ... to work with new synset numbers

0.73    Tue Dec 18 2001

    - Changes synset write() behaviour to store offsets correctly, and 
      changed Lingua::Wordnet typos. Thanks Marco Nova.

0.74    Wed Aug 11 2004

    - Fixed tests for Wordnet 2.0
    - Significant updates thanks to Markus Koppenberger and his group at 
      Universidad Pompeu Fabra:

      * Wordnet.pm
        - morph() now returns an array (it's possible that a word has more than 
          one baseform
        - reverse_morph() does the same ...
        - lookup_synset returns the synsets for all forms of morph
        - the 'functions' are commented now, because WN 2.0 uses the '+' pointer
          for derivational morphology
        - added functions for domains and domainterms (all_domains, 
          category_domains, region_domains, usage_domains and all_domainterms, 
          category_domainterms, region_domainterms,  usage_domainterms). there
          are also example 2 scripts for that functions
      
      * convertdb.pl
        - some code-cleanup: added -w and 'use strict', Removed some global 
          variables and corrected some typos. Added some comments.
        - loops over all pos-types (index.<pos>, <pos>.exc and data.<pos>
        - changed structure of lingua_wordnet.morph: if there are more than one 
          baseform, the value of the morph-hash will contain both seperated by 
          $DELIM

      * Analysis.pm
        - added a function 'global_distance', which tries to find the steps 
          needed from one to another synset. (highly experimental, very slow)

      * scripts/
        - added scripts to show the new functionalities

TODO:
    - find someone to maintain the module
    - methods for XML delivery of data