OMW Version 1

This page provides access to wordnets in a variety of languages, all linked to the Princeton Wordnet of English (PWN). The individual wordnets have been made by many different projects and vary greatly in size and accuracy. This page has (i) extracted and normalized the data, (ii) linked it to Princeton WordNet 3.0 and (iii) put it in one place. This page only includes those with a license that allows redistribution (Bond and Paik, 2012).

Search in OMW 1.3 (human curated)

We extended this with automatically extracted data for over 150 languages from Wiktionary and the ‎Unicode Common Locale Data Repository (Bond and Foster, 2013).

Search in Extended OMW 1.3 (with Wiktionary and UCLDR data)

Data and Code

• The data files are available in the OMW Data Repository
• Disambiguated Swadesh lists are available here
• Code for the wordnet browser is available here
• You can use the OMW data in the NLTK wordnet interface; you have to download the data, e.g.

  	nltk.download("wordnet")
  	nltk.download("omw-1.4")
  	# if you want the wiktionary data
  	nltk.download("extended_omw")
        

Summary of Wordnets

Language codes are linked to the English Wikipedia.

Documentation and Notes

Core

Synsets marked with ✪ are in the semi-automatically compiled list of 5000 core word senses in Princeton WordNet (approximately the 5000 most frequently used word senses). They are marked with ✪ in the interface. The original list is here from http://wordnetcode.princeton.edu/standoff-files/core-wordnet.txt (Boyd-Graber et al., 2008). Our version (converted to use collaborative interlingual index).

The wordnets are linked to the Suggested Upper Merged Ontology (Sumo: Niles and Pease, 2001; Pease, 2011); the TempoWordNet (Dias et al., 2014); the Multilingual, layered sentiment lexicons (ML-SentiCon: Cruz et al., 2014); and SentiWordNet3.0 (Baccianella et al., 2010).

Mapping between wordnet versions was done using the mappings from TALP at UPC (Daudé et al. 2000).

Formats

Tab files

The wn-data-*.tab files are tab separated files of synset-lemma pairs; or synset-subid-definition/example

# name␉lang␉url␉license
offset-pos␉lang:lemma␉word
offset-pos␉lang:def␉sid␉definition
offset-pos␉lang:exe␉sid␉example
...

Example:

# Wordnet Bahasa	ind	http://wn-msa.sourceforge.net/	MIT 
00019613-n	ind:def 0 masalah fisik yang nyata
00019613-n	ind:lemma	inti
00019613-n	ind:lemma	unsur
11407591-n	ind:def	0	Novelis dan kritikus Perancis
11407591-n	ind:def	1	pembela Dreyfus
11407591-n	ind:lemma	Emile Zola
11407591-n	ind:lemma	Zola

For this data to be really useful you need to combine it with the synset relations from the Princeton wordnet.

Known Problems

• We discard any synsets not linked to PWN (such as new synsets in the Arabic wordnet). Use OMW version2 for these.
• If the wordnet has a different structure, we only show those concepts with synonymous or near synonymous links to PWN. So for Danish, Polish and Norwegian, we only have a small subset of the entire wordnet.
• We currently only make use of synset level sentiment analysis from ML-SentiCon (Cruz et al., 2014), we do not show the language specific lemma level analysis.
• We currently can't add wordnets that don't link to PWN (such as Gaelic).
• We are focused on adding lemmas, we do not have all extra information from other projects such as:
  • Definitions and examples from wordnets such as Spanish
  • Orthographic variation and pronunciation in the Hebrew Wordnet
  We plan to add this information as time permits.
• We should strip diacritics from the Arabic wordnet to make it easier for lookup.
• We may yet be missing some available wordnets: please help us add more. Any wordnet with an open license that links to the Princeton Wordnet is welcome.
• The interface is not very multilingual.

Notes

There are some places where I made changes to harmonize different wordnets:

• Inter-word separator changed to space ' ' (some wordnets had '_'). Note that non-segmenting languages (e.g. Japanese and Thai have no inter word separator)
• Ignore the focal/satellite adjective distinction as not all wordnets make it and we don't show it in the interface. That is, map map all 's' to 'a'

References

Francis Bond and Kyonghee Paik (2012)

A survey of wordnets and their licenses In Proceedings of the 6th Global WordNet Conference (GWC 2012). Matsue. 64–71

Francis Bond and Ryan Foster (2013)

Linking and extending an open multilingual wordnet. In 51st Annual Meeting of the Association for Computational Linguistics: ACL-2013. Sofia. 1352–1362

Boyd-Graber, J., Fellbaum, C., Osherson, D., and Schapire, R. (2006): core

Adding dense, weighted connections to WordNet. In: Proceedings of the Third Global WordNet Meeting, Jeju Island, Korea, January 2006

Baccianella, Andrea Esuli Stefano and Sebastiani, Fabrizio, (2010): sentiwn

SentiWordNet 3.0: An Enhanced Lexical Resource for Sentiment Analysis and Opinion Mining., Proceedings of the Seventh conference on International Language Resources and Evaluation (LREC'10) , Valletta, Malta, 2010

Cruz, Fermín L., José A. Troyano, Beatriz Pontes, F. Javier Ortega, (2014): ml-senticon

Building layered, multilingual sentiment lexicons at synset and lemma levels, Expert Systems with Applications , 2014

Jordi Daudé, Lluís Padró and German Rigau (2000): mapp

Mapping WordNets Using Structural Information. 38th Annual Meeting of the Association for Computational Linguistics (ACL'2000), Hong Kong

Adam Pease (2011): sumo

Ontology: A Practical Guide. Articulate Software Press, Angwin, CA. ISBN 978-1-889455-10-5.

Niles, I and Adam Pease (2001): sumo

Toward a Standard Upper Ontology. In Proceedings of the 2nd International Conference on Formal Ontology in Information Systems (FOIS-2001), Chris Welty and Barry Smith, eds.

Gaël Dias, Mohammed Hasanuzzaman, Stéphane Ferrari, Yann Mathet (2014): tempo

TempoWordNet for Sentence Time Tagging. Proceedings of the Companion Publication of the 23rd International Conference on World Wide Web Companion pages 833–838, Switzerland