Agri-Ontology: A Novel Ontology Framework for Agricultural Domain

The agricultural information system deals with massive amounts of data from heterogeneous sources. It helps the farmers gain accurate information by providing better insights. A significant issue in agricultural data is its unstructured form, which can be resolved through ontology—a novel framework known as AgriOntology is proposed to address this issue for the Agrovoc dataset. The data in the unstructured form is initially pre-processed through the Natural Language Processing approach to generate features for ontology construction. The BERT and Jaccard similarity are applied to establish a relationship between the features used to construct the Agri-Ontology. Post ontology construction, the BiGAN framework is used to analyze the performance. The performance of the constructed ontology is observed to have an accuracy of 94.64%, which is higher than the existing models.