In a database, the structure of the data is described in the so-called conceptual model, which relies itself on a formalism, such as relational algebra or object classes. The model is designed so that queries on the data can be expressed to answer questions in the context of the problematics that motivated the building of the database. It is kept as simple as possible regarding these needs, essentially for the sake of efficiency.
For example, in a relational model, a gene could be described in the relation gene through several fields, such as the name of the organism, its location on the genome sequence, various annotations on its products, and so on. What is intended by gene in the context of this specific database is not made explicit, at least not formally.
When the model is extended and provides additional information that are not strictly required for storing the data and evaluating these queries, the database evolves in the direction of a knowledge base. Following the previous example, a knowledge base could explicitly specify that a gene in its context is intended to contain the coding sequences and the regulatory regions.