Understanding and using the meaning of statements in a bio-ontology: recasting the Gene Ontology in OWL

In this paper, we explain the role of a Knowledge Representation (KR) language's semantics. To illustrate the utility of language semantics we will use it to explore the reconciliation of the representations used for the Gene Ontology (GO) [1] and that used for the ontologies represented in the W3C recommendation Web Ontology Language (OWL [2]). A language's semantics is often a great concern to computer scientists, a concern that is sometimes lost on biologists. The goal of this paper is, therefore, to explain the role of language semantics to a community outside computer science (this albeit anecdotal evidence is built up over many years of teaching and tutorials in this domain between the two disciplines). In the text of this document Boldface font is used to refer to objects and logical keywords within an ontology and Italics Boldface font for terms that have a definition available in the glossary (see Additional file 1).Different knowledge representation languages provide different means to make statements about the knowledge to be captured in different ways. The semantics of these languages tell both humans and computers how to interpret statements made in those languages. Different languages have varying expressivity and computational properties, hence the corresponding tools can offer different querying and reasoning mechanisms; consequently there is often a need to exchange between languages to take advantage of their characteristics. For example, the Web Ontology Language OWL-DL [3] comes with rather high expressivity and some powerful reasoning services. As a consequence, we can annotate data using terms (and expressions built from these terms) whose meaning is defined in some OWL-DL knowledge base, usually called an "ontology", and then use a software application called a reasoner to query that data. The reasoner will take into account the definitions of the terms when answering queries, thereby providing flexible access to that data. When translating a