RDF Model Generation for Unstructured Dengue Patients' Clinical and Pathological Data

Abstract

The automatic extraction of triplets from unstructured patient records and transforming them into resource description framework (RDF) models has remained a huge challenge so far, and would provide significant benefit to potential applications like knowledge discovery, machine interoperability, and ontology design in the health care domain. This article describes an approach that extracts semantics (triplets) from dengue patient case-sheets and clinical reports and transforms them into an RDF model. A Text2Ontology framework is used for extracting relations from text and was found to have limited capability. The TypedDependency parsing-based algorithm is designed for extracting RDF facts from patients' case-sheets and subsequent conversion into RDF models. A mapping-driven semantifying approach is also designed for mapping clinical details extracted from patients' reports to its corresponding triplet components and subsequent RDF model generations. The exhaustiveness of the RDF models generated are measured based on the number of axioms generated with respect to the facts available.