Integration of Healthcare Ontologies at Schema Level using Customized Metadata

Abstract

In today’s fast growing competitive world, Data mining has become a research area of great interest as the problem of handling data in many circumstances toss lot of opportunities for research discoveries. Data being generated every second particularly in healthcare sector need to be managed efficiently so that further perusal when needed will be easier for medical professionals and researchers as an aid of decision support. Heterogeneity in the structure of data rather than the semantic discovery is the key of open challenge remained yet unaddressed. Structural construct deals at schema level of data depiction. Ontologies as means of data representation in the form of knowledge graphs are serving the field of Machine Learning (ML) from decades supporting automated knowledge extraction. Lot of research contributions are found to handle general formats to certain extent, but handling images and Portable Document Format (PDF) remain open as a major problem statement to be addressed in-order to enjoy successful information retrieval benefits. However not all relevant data is being retrieved during semantic queries due to non-homogeneity in data representation at the schema level resulting in ruling out of the document matches. In order to address the stated issue, an approach has been presented in the paper which aims at extracting metadata about the documents facing problem of heterogeneity, constructing ontologies based on the customized metadata tags followed with integration of ontologies for enhancing the prediction accuracy by increasing the relativity of documents in the semantic context. The proposed methodology can be evaluated using any of the classification techniques and solutions proved worth can be retained for daily access of semantic information thereby achieving good prediction accuracy in the process of efficient knowledge recovery.