Improving Location Reasoning with an Ontology

One characteristic of reasoning about location is the need to make use of noisy data with a range of possible interpretations. This paper presents four different algorithms that use a location ontology with ONCOR to resolve a value for location information from a range of potentially noisy sensor data, and compare their performance to three other more conventionally used algorithms. Our work takes a radical shift towards making the ontology a pragmatic tool for more than interoperability in UbiComp systems. We report an experiment that logged five people’s locations over three weeks. This enabled us to evaluate the improvement made by introducing an ontology into the location resolution process. The results show that some ONCOR algorithms yield an improvement of 50–63% on average for resolving the finest-grained location values over the Point algorithm.