Abstract
Tackling the information retrieval gap between non-technical databaseend-users and those with the knowledge of formal query languages has been aninteresting area of data management and analytics research. The use of naturallanguage interfaces to query information from databases offers the opportunityto bridge the communication challenges between end-users and systems that useformal query languages. Previous research efforts mainly focused on developingstructured query interfaces to relational databases. However, the evolution ofunstructured big data such as text, images, and video has exposed thelimitations of traditional structured query interfaces. While the existing websearch tools prove the popularity and usability of natural language query, theyreturn complete documents and web pages instead of focused query responses andare not applicable to database systems. This paper reports our study on thedesign and development of a natural language query interface to a backendrelational database. The novelty in the study lies in defining a graph databaseas a middle layer to store necessary metadata needed to transform a naturallanguage query into structured query language that can be executed on backenddatabases. We implemented and evaluated our approach using a restaurantdataset. The translation results for some sample queries yielded a 90% accuracyrate.