WordNet Mapping Scripts
Do you have tagged text, rules, or other data containing
WordNet 1.5 synsets that need
to be upgraded to 1.6? Although 1.6 comes with a mapping file for the nouns
to help you with this task, it does not come with a mapping for the verbs.
The scripts below were developed
here to automatically generate mapping files for the verbs. Naturally,
the mapping is not 100% correct, but it's close and I will update this page as
improvements are made.
Feel free to download the source and output files below. If you plan to run
the Perl scripts yourself, read through them first to see what paths you need
to change. You may modify the source, of course, perhaps to change the output
format or improve the results. Please let
me know if you do improve the results or have other suggestions.
What's New
1999-02-15 Included 1.6 to 1.5 Inverse Mapping
- Although I wrote these scripts to upgrade old senses to new senses,
it has been brought to my attention that the inverse conversion is also useful
(especially to those working with EuroWordNet data).
- output of the inverse mapping.
- wnmap1999-02-15.tar.gz
contains all the latest files.
1999-01-24 Original Release
- map is the main synset mapping script
(output).
- sense-changes is the verb sense mapping script,
which converts the output of map to verb senses
(output).
Shift-click to download.
How It Works
For each old synset, potential matching new synsets are found and ranked by the following:
- Number of synonyms that are the same
- Number of gloss words that are the same
- Number of parents that match
- Closeness of relative synset IDs
See the source for details.
To Do
- Evaluate the results
- Modify/add hueristics
- Improve documentation
Send comments or questions to
bmartin@cs.ucf.edu
.