WSD based Ontology Learning from Unstructured Text using Transformer

Representation of knowledge and making it machine comprehensible has become a necessity in modern times but with the large amount of data being generated nowadays, this process has to be automated as much as possible. In this work, we propose a deep-learning based model to build an RDF based Ontology from Unstructured Text. We aim to evaluate the proposed model by creating a general knowledge ontology from newspaper article corpora. The proposed model is based on transformer, Natural Language Processing and contains a Relation Extraction model and novel implementation of RDF mapping algorithm. The main highlight of our model is its ability to handle the Word Sense Disambiguation problem. The model was able to perform well and achieved very high accuracy scores.