How to add a wordnet to the GWG
This is preliminary page describing how to add a wordnet to the global wordnet grid.
- Send a URL pointing to a well formed LMF version of a wordnet
to gwgcili@gmail.com and
asked to be added. Someone from your project should be a member of
the GWA. Currently we have no formal requirements for deciding who
to add: the ILI maintainers (Bond, McCrae, Vossen and Fellbaum) decide.
(in the future, you will upload it somewhere instead of email)
- We will download it and:
- Validate it
if it is invalid, we will produce an error report
- Check the content for some common problems (like duplicate lemmas)
if we find any, we will produce an error report but we
can still load it
- Merge it with the ILI and other wordnets to produce the Global Wordnet Grid
- Also known as the Open Multilingual Wordnet
- All entries will be linked to the project (or projects)
that produced them
- Identify new ILI candidates
- Produce a single downloadable linked multilingual wordnet
Wordnet formatting requirements
The wordnet should be formatted as valid Simple) wordnet-LMF, using
the DTD provided.
MetaData
This should include:
project name
short project name (up to six of [-a-z])
license
url
version
release date
Valid licenses
The licence must be open according to
the open definition.
- Preferred licenses are:
- Creative Commons Attribution
- Creative Commons Attribution Share-Alike
Requirement to be added to the ILI
- The synset in your wordnet has a valid definition in English
- The definition is written according to the guidelines and is not
too similar to an existing definition.
- The synset is linked to a synset linked to the ILI
by an acceptable link
- synonym, hypernym, antonym or meronym
- It has a confidence score of 100 (or no confidence given)
- It is not a duplicate of a concept already in the ILI
- Checked automatically
- No one objects during a trail period (three months)
Other changes to the ILI
There are three other ways to change the ILI
- Edit definition
- Deprecate entry
- Supersede entry (deprecate and link to replacement)
These will have their own interface and guidelines.
More details will be given at the GWC 2016 workshop and in later
iterations of this page. This page will move to under GWA pages when
it is more polished.
Francis Bond
<bond@ieee.org>
Computational Linguistics Laboratory
Division of Linguistics and Multilingual Studies
Nanyang Technological University