Knowledge Graph Based Hard Drive Failure Prediction

Area of Science:

• Computer Science
• Data Science
• Reliability Engineering

Background:

• Hard drive failure can cause system failure and data loss, highlighting the importance of reliability.
• Existing hard drive failure prediction methods predominantly use machine learning (ML) or semantic technology independently.
• ML methods lack context-awareness, while semantic technology methods do not leverage ML's predictive capabilities.

Purpose of the Study:

• To develop an improved hard drive failure prediction method.
• To integrate the strengths of both machine learning and semantic technology.
• To enhance the accuracy and context-awareness of hard drive failure predictions.

Main Methods:

• Development of a novel knowledge graph-based approach for hard drive failure prediction.
• Leveraging ontologies and knowledge graphs (KGs) for context-awareness.
• Integrating machine learning algorithms for pattern recognition and prediction.

Main Results:

• The proposed knowledge graph-based method demonstrates superior performance in hard drive failure prediction.
• Achieved higher accuracy compared to existing state-of-the-art methods.
• Successfully integrated context-awareness with predictive capabilities.

Conclusions:

• Combining machine learning and semantic technology offers a more effective approach to hard drive failure prediction.
• Knowledge graphs provide essential context, enhancing ML-based predictions.
• The proposed method significantly improves system reliability and data loss prevention.