A Proposed Representation Framework for Semantic Science

This work addresses the problem of enabling machines to perform scientific tasks, e.g. reasoning based on scientific laws and definitions, recognizing inter-dependence of scientific domains, and answering queries about science corpus. The building blocks of science, such as scientific terms, laws, problems, solutions, theories and disciplines are traditionally represented as single, atomic nodes in scientific ontologies. This makes it difficult to distinguish those constituents and use them properly in the automation of scientific activities. We support the idea of adding structure to the representation of different constituents of science corpus. The structure of a scientific law, for instance, would be different from that of a solution to a given scientific problem. It is shown through examples that considering those different structures can help in reasoning about scientific knowledge. Moreover, the domain-independent aspects of different constituents of science have the potential to be factored out in a meta-ontology. This meta-science can also contain general reasoning machinery about science.