WEIGHTED ENTITY-LINKING AND INTEGRATION ALGORITHM FOR MEDICAL KNOWLEDGE GRAPH GENERATION

Abstract

Semantic data integration is the process of interrelating information from multiple heterogenous resources. There is a need for representation of data concepts and their relationships to eliminate heterogeneity among different data sources in healthcare management systems. Standardized medical ontologies provide predefined medical vocabulary serving as a stable interface for concepts related to medical data sources. However, different ontologies have different concepts although these concepts have logical relations between them such as the Human Disease Ontology (DO) and the Symptoms ontology (SYMP). There is a need for a knowledge graph providing a reliable knowledge base for any intelligent healthcare expert advisor disease prediction systems. The knowledge graph provides a model for linking and integrating different concepts having logical relationships such as diseases and their symptoms. Medical online website and encyclopedia provides a reliable source for building such a knowledge graph. The knowledge graph is enriched with social networks data where information extracted reflects a major source of data based on user experiences. The paper proposes a framework for constructing a disease-symptom entity linked integrated knowledge graph based on online medical encyclopedia and social networks user experiences. Entity linking such knowledge graph with standardized medical ontologies makes it a reliable knowledgebase for a standard system that could be used by social networks user and the professional staff.