An Ontology Learning Framework for unstructured Arabic Text
Mohammad Mustafa Taye, Rawan Abulail, Mohammad Aloudat · 2023
Ontologies are widely regarded as valuable sources of semantics and interoperability in all artificially intelligent systems. Due to the rapid growth of unstructured data on the web, studying how to automatically get ontology from unstructured text is important. Therefore, ontology learning (OL) is an important process in the business world. It involves finding and extracting concepts from the text so that these concepts can be used for things such as information retrieval. Unfortunately, learning ontology is not easy for some reasons, and there has not been much research on how to automatically learn a domain-specific ontology from data.Ontology Studying Arabic text is not as developed as learning Latin text. There is almost no automated support for using Arabic literary knowledge in semantically enabled systems. Machine learning (ML) has proven beneficial in numerous fields, including text mining. By employing neural language models such as AraBERT, it is possible to obtain word embeddings as distributed word representations from textual input using machine learning. However, the application of machine learning to aid the development of Arabic ontology is largely unexplored. This research examines the performance of AraBERT for ontology learning tasks in Arabic. Early performance results as an application of Arabic ontology learning are promising. In this research, we provide a method for populating an existing ontology with instance information extracted from the input natural language text. This prototype has achieved an information extraction accuracy of 91%.