LexEQUAL: supporting multilexical queries in SQL

Abstract

Current database systems offer support for storing multilingual data, but are not capable of querying across languages, an important consideration in today's global economy. We therefore propose a new multilexical operator called LexEQUAL that extends the standard lexicographic matching in database systems to matching of text data across languages, specifically for names, which form close to twenty percent of text corpora. The implementation of the LexEQUAL operator is based on transforming matches in language space into parameterized approximate matches in the equivalent phoneme space. A detailed evaluation of our approach on a real data set shows that there exist settings of the algorithm parameters with which it is possible to achieve both good recall and precision.