Related Experiment Video
Updated: Dec 23, 2025

Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
Published on: June 13, 2025
Bootstrapping Adversarial Learning of Biomedical Ontology Alignments
Ramon M Maldonado1, Sanda M Harabagiu1
1University of Texas at Dallas, Richardson, TX, U.S.A.
This study introduces KAEGAN, a novel approach for aligning biomedical ontologies using knowledge graph embeddings. Jointly learning alignment and representation enhances performance over isolated methods.
Area of Science:
- Biomedical Informatics
- Artificial Intelligence
- Knowledge Representation
Background:
- Automatic alignment of biomedical ontologies is crucial due to their increasing complexity and widespread use.
- Neural learning techniques can leverage knowledge graph embeddings for ontology representation and alignment.
Purpose of the Study:
- To present the Knowledge-graph Alignment & Embedding Generative Adversarial Network (KAEGAN).
- To demonstrate KAEGAN's capability in representing relational knowledge and aligning biomedical ontologies using semantics.
Main Methods:
- KAEGAN utilizes a Generative Adversarial Network (GAN) framework.
- The model is trained using bootstrapping for iterative alignment improvement.
- It learns knowledge embeddings from distinct biomedical ontologies.
Main Results:
- Experimental results indicate promising performance for KAEGAN.
- Jointly learning ontology alignment and knowledge representation yields superior outcomes.
- The approach effectively utilizes ontology semantics for alignment.
Conclusions:
- KAEGAN offers a promising method for automated biomedical ontology alignment.
- Integrating knowledge representation with alignment learning enhances overall effectiveness.
- This work advances the field of biomedical knowledge graph integration.
Related Concept Videos
Improving Translational Accuracy
Improving Translational Accuracy
Associative Learning
Classical conditioning, also known...
Genome Annotation and Assembly
Evolutionary Relationships through Genome Comparisons
Introduction to Learning
In contrast to learned behaviors, unlearned behaviors such as crying, sexual...

