NLP to SQL : TRANSLATING NATURAL LANGUAGE TO DATABASE QUERY LANGUAGE

Abstract

Using Query language to fetch and visualize the data in specified format requires knowledge and expertise about domain and technologies in which data are being stored in databases. The gap between what information user wants which is going to satisfies the need at the desired moment and what information is being fetched requires clear communication between database expert and user requesting the data. In this project I understudy the different methodologies which was proposed in the domain of natural language to query language such as Seq2SQL, Neural Enquirer, ONLI (Ontology-based Natural Language Interface), NaLIR (Natural Language Interface for Relational Database), SQLNet and ln2sql put the comparison and features of databases which are being used to train the model. Training Datasets are WikiSQL, Spider, DBpedia, Geoquery, ATIS, Overnight, WebQuestions, Freebase917 and WikiTableQuestions. The main challenges of data retrieval are the heterogenous form of data storage which requires complex and different domain specific queries. In extension of this work, I propose how complex queries can be fetched from multiple normalized database tables.