Automatic detector of term pairs

Lex2 is a tool for automatically creating dictionaries. Using parallel corpora (with translation memories), it detects and automatically extracts term pairs from correspondences in Basque and Spanish; or, in other words, the tool extracts bilingual lexicons.

It uses linguistic and statistical methods: it identifies the term candidates in the text in each language and, on the basis of frequency and other types of statistical information, it establishes possible links between the candidates of both languages. When said links are likely to be sufficiently significant, it lines up the candidates in order to create a bilingual lexicon.

Consequently, the search process is not the same as that used by the iTXEK checker.   Lex2 does not seek the correspondences that a term in the source language may have in the target language. In fact, the main purpose of Lex2 is not to check for the proper or incorrect use of terminology. Lex2 separately seeks and extracts the term candidates from each language and, going one step further, seeks statistical parallelisms in order to detect links between candidates. It is thus capable of extracting bilingual lexicons from the terms used in large text masses.