Semantic Science: machine understandable scientific theories and data
David Poole · 2007
The aim of semantic science is to have scientific data and scientific theories in machine understandable form. Scientific theories make predictions on data. In the semantic science future, whenever someone does a scientific experiment, they publish the data using a formal ontology so that there is semantic interoperability; it can be compared with other data collected by others, and used to compare theories that make prediction on this data. When someone publishes a new theory, they publish it with respect to an ontology so they can test it on all available data about which it makes predictions. We could all see which theories predict the data better. By the use of formal ontologies, we could determine which are competing theories (when they make different predictions for the same data) and which are complementary. Whenever new data is collected, we can determine which theory better predicts the data. Human-made scientific theories can be compared with machine learned theories (of course, most theories are a mix). Imagine now the best theories applied to new cases: we can use the best medical theory to predict the disease a patient has, the best geological theory to predict where landslides will occur or the best economic theory to predict the effect of a policy change. This paper is preliminary and always under construction. If you have feedback, more references, please