Code for the representation, management and application of a physics knowledge graph
Knowledge graphs represent large volumes of human knowledge using words, which excludes the representation of continuous mathematical knowledge from the sciences. A knowledge graph is introduced to represent and manage knowledge from Physics. The knowledge graph is then used for the application of equation recovery from input and output data of an unknown equation. Three methods are used for this application- a dimensional analysis, a brute force search of the knowledge graph, and embeddings that project entities of the knowledge graph into the vector space of Taylor series coefficients. These methods significantly reduce the search space for equation recovery, demonstrating the advantages of knowledge graphs in equation recovery.