Isolating significant phrases in common natural language queries to databases

Abstract

This paper describes a novel methodology for extracting significant phrases from common natural language queries to databases. We show the usability of the methodology for developing database front ends, and propose it as an acceptable alternative to developers of databases that are supposed to be subject to massive interaction by non-technical people. Here we localize the matching procedures by isolating the phrases that are important for mapping the queries to database details, and propose to use heuristics and soft computing techniques for inexact matching of the phrases.