A Vocabulary of Topological and Containment Relations for a Practical Biological Ontology

We describe the development of a formal language for expressing qualitative spatial knowledge. The language is intended as a practical tool for knowledge representation and has been designed with the particular aim of encoding the qualitative spatial information found in an introductory college level biology textbook. We have taken a corpus-driven approach in which we first identify the requirements by analysing the sentences in the book, design the vocabulary, and then check its adequacy by applying it to model the sentences containing spatial knowledge. Our technical solution extends the well-known Region Connection Calculus with predicates for referring to surfaces, cavities and different forms of containment. We illustrate the application of this vocabulary in encoding sample sentences from the book, and give empirical results regarding the correspondence between the defined relations of our formal theory and the actual usage of vocabulary pertaining to ‘surrounding’, ‘enclosure’ and ‘containment’ in the textbook.