Welcome to the Collaborative Interlingual Index
The Collaborative Inter-Lingual Index (CILI: pronounced like the chili pepper) is a registery of concepts, used by wordnets in the Open Multilingual Wordnet to link meaning across languages.
In simple terms, CILI is a list of ID’s corresponding to concepts, linked to the wordnet that first defined it. Each unique concept has a unique ID (called ILI ID), and a definition in English. This will link to synsets, and senses in many languages, e.g. January, Januar, Janvier (or January in some other language), all will have the same ILI ID (i117152 in this case) and definition.
Motivation behind CILI
Issues like the inclusion of similar or even identical concepts were arising while linking concepts across languages. There was a requirement of something which brings coordination between different wordnets and eases the sharing of concepts among the wordnet builders. A registry of concepts was purposed as a solution to these issues. Consequently, CILI got developed to resolve these issues. The structure of the CILI is based on the Interlingual Index first proposed in the EuroWordNet project.
Back to the topBrief Overview of CILI
CILI is released under a Creative Commons Attribution 4.0 (CC BY) license. It has definitions in English and links to wordnets in the Open Multilingual Wordnet (OMW). The CILI is a collection of concepts to which all wordnets are linked. It does not duplicate the relations between these concepts as represented in any wordnet, and it does not have any lexicalizations.
Its simple design allows us to link wordnets with a minimum of extra work. Once concepts got added to the CILI, they will get a persistent ID and thereafter should not be deleted or change in meaning. We make a simple requirement that definitions have a minimum length of twenty characters or five words.
Back to the topProperties of the CILI
- 1. The Collaborative Interlingual Index (CILI) should be a flat list of concepts.
- 2. The semantic and lexical relations should
mean the same things for all languages.
⇒ documentation for relations - 3. Concepts should be constructed for salient and frequent lexicalized concepts in all languages.
- 4. Concepts linked to multiword expressions (MWEs) in wordnets should be included.
- 5. A formal ontology could be linked to but separate from the wordnets.
- 6. The license must allow redistribution of the index.
- 7. CILI IDs should be persistent: we never delete, only deprecate or supercede, we should not change the meaning of the concept.
- 8. Each new CILI concept should have a definition in English, as this is the only practical way we can coordinate across languages. The definition should be unique. Definition changes will be moderated.
- 9. Each new CILI concept should link to a synset in an existing project that is part of the OMW with one of a set of known relations (hypernymy, meronomy, antonymy, ...).
- 10. This synset should link to another synset in an
existing project that is part of the OMW and
links to a CILI concept:
⇒ each concept is linked to another concept through at least one wordnet in the OMW. - 11. New concepts can be added if they exist in an uploaded wordnet:
- Any project adding new synsets should first check that they do not already exist in the CILI;
- If something fulfills the above criteria it can be proposed;
- If there are no objections after three months then it is accepted as part of CILI.
Refer Toward a truly multilingual Global Wordnet Grid (Francis Bond, Piek Vossen and John McCrae, 2016) for more information.