Laying Lexical Foundations for NLP: the Case of Basque at the Ixa Research Group

The purpose of this paper is to present the strategy and methodology followed at the Ixa NLP Group of the University of The Basque Country in laying the lexical foundations for language processing. Monolingual and bilingual dictionaries, text corpora, and linguists’ knowledge have been the main information sources from which lexical knowledge currently present in our NLP system has been acquired. The main lexical resource we use in research and applications is a lexical database, EDBL, that currently contains more than 80,000 entries richly coded with the lexical information needed in language processing tasks. A Basque wordnet has also been built (it has currently more than 50,000 word senses), although it is not yet fully integrated into the processing chain as EDBL is. Monolingual dictionaries have been exploited in order to obtain knowledge that is currently being integrated into a lexical knowledge base (EEBL). This knowledge base is being connected to the lexical database and to the wordnet. Feedback obtained from users of the first language technology practical application produced by the research group, i.e. a spelling checker, has also been an important source of lexical knowledge that has permitted to improve, correct and update the lexical database. In the paper, doctorate research work on the lexicon finished or in progress at the group is outlined as well, as long as a brief description of the end-user applications produced so far.