GLAWI, a free XML-encoded Machine-Readable Dictionary built from the French Wiktionary

Authors: Franck Sajous, Nabil Hathout

This article introduces GLAWI, a large XML-encoded machine-readable dictionary automatically extracted from Wiktionnaire, the French edition of Wiktionary. GLAWI contains 1,341,410 articles and is released under a free license. Besides the size of its headword list, GLAWI inherits from Wiktionnaire its original macrostructure and the richness of its lexicographic descriptions: articles contain etymologies, definitions, usage examples, inflectional paradigms, lexical relations and phonemic transcriptions. The paper first gives some insights on the nature and content of Wiktionnaire, with a particular focus on its encoding format, before presenting our approach, the standardization of its microstructure and the conversion into XML. First intended to meet NLP needs, GLAWI has been used to create a number of customized lexicons dedicated to specific uses including linguistic description and psycholinguistics. The main one is GLÀFF, a large inflectional and phonological lexicon of French. We show that many more specific on demand lexicons can be easily derived from the large body of lexical knowledge encoded in GLAWI.

Keywords: French Machine-Readable Dictionary; Free Lexical Resource; Wiktionary; Wiktionnaire

Reference: In Kosem, I., Jakubiček, M., Kallas, J., Krek, S. (eds.) Electronic lexicography in the 21st century: linking lexical data in the digital age. Proceedings of the eLex 2015 conference, 11-13 August 2015, Herstmonceux Castle, United Kingdom. Ljubljana/Brighton: Trojina, Institute for Applied Slovene Studies/Lexical Computing Ltd., pp. 405-426.


Published: 2015