Dealing with decisional natural language why-question in business intelligence

Abstract

Business Intelligence aims at providing information to non expert users to make appropriate decisions. To access decisional information, the decision maker should express his/her requirements in Natural Language interfaces instead of technical languages, to avoid IT designer dependency and formal language's unfamiliarity. A Natural Language query is formulated according to several types such as WH-questions as “What, Who, Where, etc.” questions, Keyword-like questions, etc. We focus on “Why-Question” and we present an approach handling this kind of question. We begin by modelling the Why-Question, then we present a mathematical model based on numerical analysis to detect the eventual Why-Questions answers. We propose a tool called Why-Question analyser implementing our approach. A preliminary evaluation is presented.