Enhancing Biomedical Knowledge Representation Through Knowledge Graphs

Abstract

There is a plethora of information related to the biomedical domain on the internet. Unfortunately, retrieving this information online is challenging because of insufficient semantic metadata embedded within the web documents that help search engines interpret the biomedical information. Semantic annotators have partially bridged this gap, yet these tools frequently need to catch up in accuracy, speed, and the ability to dynamically represent knowledge. We initially developed "Semantically," a biomedical semantic content authoring platform to streamline and enhance biomedical annotations through a social-technical approach. Even so, the current system stores data in a relational schema, which lacks machine-readable content that allows search engines to parse the meanings to annotation recommendations. There is still the need for the amalgamation and contextually rich representation of annotation recommendation information to enhance navigation and exploration of data. Therefore, we propose a knowledge graph-based recommendation system with an nlp-enhanced search query to provide an environment for easy and quick access to optimal recommendations in a machine-readable knowledge graph format. We obtain results for the knowledge graph through an evaluation survey that substantiates the efficacy of our knowledge graph-based recommendation system, highlighting its role in advancing dynamic knowledge representation and semantic annotation in the biomedical domain. A demo is available at SebC750/Semantically at Knowledge_Graph_Branch (github.com)

Keywords

• knowledge graph
• biomedical annotations
• knowledge representation
• ontologies
• social-technical