Design of a natural language front-end to a relational database

Abstract

The need for Natural Language Interfaces to Databases (NLIDB) has become increasing acute as more and more people access information stored in databases. However, getting better database querying skills can involve a significant investment of time especially for laymen. The purpose of this project was to research on both the theoretical and implementation of Natural Language Interfaces to Databases. It analyses how syntactic knowledge can be mapped to the corresponding SQL. Several parsing techniques have been explored with much emphasis on top down parsing using recursive descent algorithm. A system has been developed in PerI to be database and platform independent with a graphical user interface implemented using PerI Tk. The system uses Parse: :Rec Descent module to build the underlying parser, which is used to translate natural language queries to SQL statements. It has also the ability to learn new grammar. The system has been tested on library domain knowledge, and has been able to translate natural language input to SQL based on SELECT commands. Extending the grammar and translating non-select commands have been given as recommendations for future work.