A new framework for building agricultural domain-based ontologies from text documents using natural language processing and artificial intelligence techniques

Article type: Research Article

Authors: Saravanan, Krithikha Sanju^{; *} | Bhagavathiappan, Velammal

Affiliations: Department of Computer Science and Engineering, College of Engineering, Guindy, Anna University, Chennai, India

Correspondence: [*] Corresponding author. Krithikha Sanju Saravanan. E-mail: [email protected].

Abstract: The advancements in technology, particularly in the field of Natural Language Processing (NLP) and Artificial Intelligence (AI) can be advantageous for the agricultural sector to enhance the yield. Establishing an agricultural ontology as part of the development would spur the expansion of cross-domain agriculture. Semantic and syntactic knowledge of the domain data is required for building such a domain-based ontology. To process the data from text documents, a standard technique with syntactic and semantic features are needed because the availability of pre-determined agricultural domain-based data is insufficient. In this research work, an Agricultural Ontologies Construction framework (AOC) is proposed for creating the agricultural domain ontology from text documents using NLP techniques with Robustly Optimized BERT Approach (RoBERTa) model and Graph Convolutional Network (GCN). The anaphora present in the documents are resolved to produce precise ontology from the input data. In the proposed AOC work, the domain terms are extracted using the RoBERTa model with Regular Expressions (RE) and the relationships between the domain terms are retrieved by utilizing the GCN with RE. When compared to other current systems, the efficacy of the proposed AOC method achieves an exceptional result, with precision and recall of 99.6% and 99.1% respectively.

Keywords: Anaphora resolution, term extraction, relationships identification, RoBERTa model, regular expressions, graph convolutional network, domain ontology

DOI: 10.3233/JIFS-237632

Journal: Journal of Intelligent & Fuzzy Systems, vol. Pre-press, no. Pre-press, pp. 1-19, 2024