A semantically enriched medical literature mining framework

In this paper we present a literature mining framework that, by using web services offered by Pubmed, automatically retrieves a set of documents, as well as information on genes, diseases and proteins and then builds a set of matrices that are classified according to a precise methodology. The classification results are explored and combined in order to obtain associations among the main feature domains: genes and diseases. The associations are represented in an RDF association graph that can be queried using a SPARQL client. The association graph is enriched with other linked data of relevant ontologies: the Go ontology, the Human Disease Ontology and the Protein Ontology in order to further extend the discovered association with related information in order to establish a network of interacting elements involved in one disease. The element of the network can be annotated using the annotations present in the GO ontology.