ROBOT: A High Performance Natural Language Data Base Query System

Abstract

The field of natural language data base query has seen several successful research systems that have been able to process large subsets of the English language. The ROBOT system is a high performance natural language processor that extends the techniques developed in these earlier systems, in an attempt to provide a usable natural language query medium for non-technical users. ROBOT is already installed in four real, world environments, working in five application areas. Analysis of log files indicate that between 80-90% of end user requests are successfully processed. The system successfully deals with both lexical and structural ambiguity, inter- and intrasentential pronomilization and sentence fragments. The resource requirements of ROBOT are well within the limits of medium to large computer systems. This paper describes the system from an information flow point of view. The goal is to obviate the creation of a semantic store purely for the natural language parser, since a different structure would be required for each domain of discourse. Instead the data base itself is used as the primary knowledge structure within the system. In this way the parser can be interfaced to other data bases with only minimal effort.