An Ontological Approach for Natural Language Command Interpretation and Its Application in Robotics

Abstract

We present an ontological approach for interpreting natural language commands as well as an application to intelligent robots. In our approach, we first extract meaningful keywords from the user commands by adopting lexico-semantic pattern matching. A structured representation is then built by extracting and arranging meta-data for each keyword. The data is then transformed into an ontological knowledge representation. We implemented a prototype interpretation system, and applied it to the voice user interface of an intelligent mobile robot. The system was tested with several commands for querying weather or traffic information and controlling the robot and various digital home devices. The present experiments with an ontological based interpretation scheme demonstrate that the coverage of user commands can be easily extended, as the representation is logical and declarative