Automated Query Processing using Natural Language

Abstract

At present, businesses have very large amount of information and users need to generate various reports frequently based on their changing requirements. Custom queries can be used to generate these reports using RDBMS software’s and it needs a prior understanding of SQL scripting including database structure. In most scenarios, these databases have more complex relationships and it is very much difficult to understand by a common user. If a user needs a new report, that person has to ask the IT department to generate and get it done. In most cases, it will take couple of hours to one or two days to get it completed. Further, SQL scripting is not something familiar among common users and it needs a bit of technical understanding as well. However, Executive system users or CEOs do not have much time to learn these things and time to type SQL scripts to generate ad-hoc reports. They need information by click of a button. This research basically focuses on basic query processing using natural language, complex query processing using natural language and voice authentication and recognition. Speaker recognition is the process of automatically recognizing who is speaking on the basis of individual information included in speech waves. This technique makes it possible to use the speaker's voice to verify their identity and access database components based on their privileges. MFCC algorithm has been implemented to facilitate voice authentication. Once the access is granted, it enables the user to retrieve records dynamically based on their privileges using natural language. The voice is captured using voice recognition techniques and then it is converted back into SQL. The conversion is facilitated by using a wrapper and it maps the natural language to SQL. Once it is converted, then it is executed against the respective database to generate the result set. The conversion facilitates select statements including relationships up to level three. The objective has been successfully achieved using MFCC algorithm for voice recognition and SQL wrapper for natural language conversion to SQL. This research has a great potential to broaden up its work and expand this capability to reach out for disabled people. Since this is capable of handling voice, the people who are unable to use a keyboard can easily be trained to get the advantage of this. Further, the current scope of this project is limited only for select statement and can be further enhanced to support insert, delete and update statements.