Comparative Study of Pipeline and Deep Learning Approach for NLIDB: Perspectives and Challenges

Abstract

Today's technology-rich world depends greatly on database, and Query languages to access database such as sql, are not common knowledge. Attribute Based Access Control (ABAC) systems are latest research trend in the field of IOT. Policy and rules are needed to get extracted from natural language documents for ABAC model. A system that can access the data via everyday language could be a huge game changer for data science and IOT. Such interface to data is focus of research for a while now, in form of Natural Language Interface to Database (NLIDB) Systems. Primary task of an NLIDB is to translate Natural language expression into structured query language. State of the art systems developed for this task adopted mainly one of 2 architectural approaches. 1- pipeline approach 2- Deep Learning approach. In this paper we conduct a comparative study of these 2 approaches. Our study displays a brief analysis of gains and research challenges for each approach.