AI Model to Generate SQL Queries from Natural Language Instructions through Voice

Abstract

Data plays the most important role in the development of industries, small businesses. Even world leaders need the data to make analyses and make better policies for people. In almost every field where the work process is digitized need to store data and then retrieve it. According to statistics most of the data is stored in the relational database and for manipulations of the data, Structured Query Language(SQL) is commonly used. So for handling databases a person need to have specialized knowledge regarding the queries and had to remember the syntax of many complex queries. So to enhance data manipulation using SQL and to efficiently get the required query, the paper proposes a method for the generation of SQL query from natural language input, spoken(audio input) by the user. The model is constructed on NLP (Natural Language Processing) and Neural Networks (Deep Learning) technologies. Long Short Term Memory(LSTM) Model is used for predicting queries and is trained on the dataset with natural language as input and returns outline skeletal structure of the query as output. Then the output will be processed and the final query will be displayed to the user. The project also aims to benefit the people who are suffering from Repetitive Stress Injury (RSI), causing pain in the finger joints, which has been attributed to work requiring a long period of typing and also to those who are not familiar with SQL queries. As this system will readily provide the required query.