Natural Language Access to Relational Databases through STEP

This paper introduces the STEP system for natural language access to relational databases. In STEP the administrator couples phrasal patterns to elementary expressions within a decidable fragment of tuple relational calculus. This phrasal lexicon serves as a bi-directional grammar, enabling the generation of natural language from tuple relational calculus and the inverse parsing of natural language to tuple calculus expressions. This ability to both understand and generate natural language enables STEP to engage the user in clarification dialogs when the parse of their query is of questionable quality. The STEP system is implemented and is currently being evaluated over a geography database. Additional evaluations are planned and will also be available on STEP’s website (http://www.cs.umu.se/ ̃mjm/step).