Mundus Traducere : Interpretation of natural language using a semantic sensor network

UNDERSTANDING natural language utterances cannot be done by studying only language. Models of the real world are also needed. Those models are usually built around semantic data. As sensor networks allow us to get information about the physical world, we wanted to add said information to the models used to interprete natural language. Adding that kind of physical data to world models will lead to dynamic model that can be deployed in everyday life systems like smart houses. This paper will focus on the theory needed to implement such physical data to semantic data and will present an implementation of previously mentionned theory called Mundus Traducere. This implementation is a Java program that generates formal expressions based on natural language utterances, linking real world objects with their language counter parts. The implementation is not regarded as a stand alone application but as a tool for implementing more complex interfaces.