A review of feature selection strategies utilizing graph data structures and Knowledge Graphs.

Area of Science:

• Computer Science
• Bioinformatics
• Artificial Intelligence

Background:

• Feature selection (FS) is crucial for Knowledge Graphs (KGs) in fields like biomedical research and NLP.
• KGs enhance machine learning (ML) model efficacy, hypothesis generation, and interpretability.

Purpose of the Study:

• To review methodologies for feature selection in Knowledge Graphs.
• To emphasize the role of FS in improving ML model performance and interpretability.
• To catalyze innovation in KG feature selection for diverse applications.

Main Methods:

• Comprehensive review of existing feature selection techniques for KGs.
• Analysis of critical factors including scalability, accuracy, and interpretability.
• Exploration of domain knowledge integration and multi-objective optimization.

Main Results:

• Identified scalability, accuracy, and interpretability as critical for KG FS.
• Highlighted the potential of integrating domain knowledge to refine FS.
• Showcased the impact of multi-objective optimization and interdisciplinary collaboration.

Conclusions:

• Feature selection is vital for advancing KG-based analytical models.
• Future directions include scalable, dynamic FS algorithms and explainable AI integration.
• KG feature selection holds transformative potential for precision medicine and other fields.