Deep Ontology Alignment Using a Natural Language Processing Approach for Automatic M2M Translation in IIoT.

Area of Science:

• Industrial Automation
• Internet of Things (IoT)
• Artificial Intelligence

Background:

• Modern Industry 4.0 and 5.0 rely on interconnected devices for optimized resource management.
• Industrial Internet of Things (IIoT) faces challenges with heterogeneous devices lacking compatible designs and interoperability.
• Conventional solutions for device integration are costly and require extensive engineering effort.

Purpose of the Study:

• To propose a self-learning model for determining device taxonomy and enabling seamless integration.
• To address the challenge of integrating new devices with different ontologies into existing IoT networks.
• To facilitate dynamic machine-to-machine (M2M) translation without additional engineering or hardware.

Main Methods:

• Utilizing a self-learning model to analyze ontological meta-data and structural information of devices.
• Employing Natural Language Processing (NLP) to match distinct ontologies based on linguistic contexts.
• Visualizing ontological networks as knowledge graphs to understand meta-data structure and message formulation.
• Aligning entities of ontological graphs with similar context and structure.

Main Results:

• The model successfully determines device taxonomy and identifies matches between different ontologies.
• It enables dynamic M2M translation, enhancing interoperability within IIoT networks.
• The approach reduces the need for manual engineering and additional hardware resources for device integration.

Conclusions:

• The proposed self-learning model offers an efficient and cost-effective solution for device interoperability in Industry 4.0/5.0.
• It facilitates dynamic translation and integration of heterogeneous devices, crucial for smart cities and industrial automation.
• This advancement supports the seamless expansion of IoT networks and optimizes energy distribution and resource management.