This article presents some structural, ontological, syntactical and lexical choices to make during knowledge representation and storage, for improving the precision, inter-connection, comparison, retrieval and re-use of knowledge objects within and between ontologies on the Semantic Web. Most of these choices (or guides for such choices) can be enforced or seen as criteria to evaluate ontologies. Some notations and methods supporting these choices are introduced. These guides are distinct from – and complementary to – those of particular knowledge acquisition methods and those advocating the use of particular top-level ontologies.