Building Lexical Resources: Towards Programmable Contributive Platforms

Abstract

Lexical resources are very important in nowadays society, with the globalization and the increase of world communication and exchanges. There are clearly identified needs, both for humans and machines. Nevertheless, very few efforts are actually done in this domain. Consequently, there is an important lack of freely available good quality resources, especially for underresourced languages. Furthermore, the majority of existing bilingual dictionaries is built with one language as English. Therefore, if one wants to translate from one language (that is not English) to another, it uses English as a pivot. And even for English native speakers, it creates a lot of misunderstandings that can be critical in many situations. In order to create and extend freely available good quality rich lexical resources for under-resourced languages online with a community of voluntary contributors, Jibiki, an online generic platform for managing (lookup, editing, import, export) any kind of lexical resources encoded in XML, has been developed. This platform is successfully used in several dictionary construction projects. Concerning the data, a serious game has been launched in order to collect precious lexical information such as collocations that will be integrated later into dictionary entries. Work is now done on extending our platform in order to reuse the resulting resources and enriching them by synchronization with the other systems (language learners and translators environments, machine translation systems, etc.).