Syntax and Table Aware Parsing Based Naturalized Structured Query Language
Praveena Mydolalu Veerappa, Ajeet Annarao Chikkamannur · International journal of intelligent engineering and systems · 2019
A database is characterized as accumulation of data that is organized to access, manage, and update information effectively.All information is store in a database and there are numerous approaches to interface with the database to access our information.A client needs some specialized knowledge to extricate information from the database.They have to utilize Structured Query Language (SQL) for information definition, information manipulation, or information control.Although, the vast majority of the clients who need to separate information from a database are not technical experts.Hence, there is a tremendous communication gap between the database and its core client.With the advancement of Natural Language Processing (NLP), a client would now be able to converse with their database in their characteristic language without learning database language.The communication gap between the client and the database has begun to vanish with this stunning capacity.In this paper, the methodology addresses this issue by considering the structure of table and the syntax of SQL.The nature of the generated SQL query is fundamentally enhanced through finding out how to recreate content from column names, cells or SQL keywords; and enhancing the generation of WHERE clause by utilizing the column-cell relation.Analyses are directed on WikiSQL, a recently released dataset with the biggest query SQL sets.Our approach significantly improves the state-of-the-art precision, recall and F-measure from 76.5% to 82%.