Inferring SQL Queries Using Interactivity

Abstract

Interactivity in language processing plays a pivotal role to allow models to better understand how to build the appropriate output. In the task of Natural Language to SQL, the fact of including the users' interactivity can be one of the practical solutions that haven't been studied deeply in the existing works published in the last decade. Using databases by users with limited familiarity in SQL will create an additional obstacle for these users to better exploit the content stored in the database systems. In this paper we present the already published studies and we discuss the utility of using the interactivity to definitely improve the query generation process in order to construct a model that generalize for unseen and complex sentences and to automatically generate the appropriate outputs.