Natural Language Interface for Databases (Machine Learning)

Abstract

Encephalon is a trainable natural language interface for databases based on template matching and artificial intelligence to convert the human language into an appropriate SQL query for relational database system. The reason for building natural language interfaces for systems as opposed to formal language systems e.g. SQL is to make interaction for naive users easier despite having a little knowledge about the system. Such front ends relieve the users of the need to know the structure of the database and offer a much more convenient and flexible interaction. This interface acts as an abstract middleware between user and database systems. Previously many interfaces like these have been developed but their analysis show that people are generally unwilling to trade reliable and predictable interfaces with intelligent but unreliable ones. While designing this system, therefore, four critical factors from clients perspective viz. extensibility, portability, domain independence and enhanced sense modality were kept under consideration and to increase the percentage of accuracy a new set of rules and templates has been introduced.