.

ISSN 2063-5346
For urgent queries please contact : +918130348310

Performance Evaluation of Automated Construction of Domain-Based Ontology Using CNN and Machine Learning

Main Article Content

Giridhar Urkude, Moon Banerjee, Bala Subramanyam P N V, B Lakshmana Swamy, Mohan Awasthy
» doi: 10.31838/ecb/2023.12.si4.201

Abstract

Ontology is essential to the Semantic web and many new AI applications. With the assistance of ontology, the client and the system may interact with the popular domain understanding in a machine-to-machine environment. While ontology has been proposed as an essential tool to reflect real-world information in database design construction, most ontological innovations are not immediately implemented. The fundamental challenges to developing and updating these Domain-specific Ontologies, such as the need for manual involvement by field experts and the limitations imposed by current technology, have made it less feasible for ontology to create and upgrade automatically. Automatic ontology generation plays therefore an essential role in semantical web applications and emerging AI applications. However, domain experts need manual involvement to help build and update the domain-specific ontology, and the limitations on existing technology adoptions make it less feasible to create and update ontologies automatically. The key contribution of this research work is to generate automated ontologies from discrete sources of knowledge based on machine learning algorithms. Study findings are in four distinct stages, data extraction from many sources, automated acquiring of information using machine learning algorithms and selection of automatic attributes, and the modelling and validation of the relationship between entities and ontology. A model is developed for acquiring and exploring information in field ontology using the Jaccard Relationship and the Neural Network. The results show that the machine-built models can construct domain-specific ontologies automatically and efficiently.

Article Details