Levels and building blocks—towards a domain granularity framework for the life sciences
The use of online data repositories and the establishment of new data standards that require data to be computer-parsable so that algorithms can reason over them have become increasingly important with the emergence of high-throughput technologies, Big Data and eScience. As a consequence, there is an increasing need for new approaches for organizing and structuring data from various sources into integrated hierarchies of levels of entities that facilitate algorithm-based approaches for data exploration, data comparison and analysis. In this paper I contrast various accounts of the level idea and resulting hierarchies published by philosophers and natural scientists with the more formal approaches of theories of granularity published by information scientists and ontology researchers. I discuss the shortcomings of the former and argue that the general theory of granularity proposed by Keet circumvents these problems and allows the integration of various different hierarchies into a domain granularity framework. I introduce the concept of general building blocks, which gives rise to a hierarchy of levels that can be formally characterized by Keet's theory. This hierarchy functions as an organizational backbone for integrating various other hierarchies that I briefly discuss, resulting in a general domain granularity framework for the life sciences. I also discuss the implicit consequences of this granularity framework for the structure of top-level categories of 'material entity' of the Basic Formal Ontology. The here suggested domain granularity framework is meant to provide the basis on which a more comprehensive information framework for the life sciences can be developed.