A multi-agent system to learn literature ontology: An experiment on English Quran corpus

This paper proposes a Multiple-Agent system to construct an ontology of the English translations of the meanings of Noble Quran. None of the current agent based systems have solved the problem of ontology learning from text documents. The Noble Quran has an unconventional style of discourse containing recurring themes and a plethora of versatile figures of speech making it an inimitable and unique literature. An array of Intelligent Agent instances help in automate the task of ontology learning. The existing vocabulary, concordances, and Index of the Quran which are readily available in electronic form are used as data sources by the multiple agents for the ontology construction process. Alsoan agent starts up with relatively small seed ontology prepared by domain experts as a starting point for the construction of a comprehensive ontology. WordNet lexicon is used to cluster the index terms utilizing the Hypernym, Holonym, Synonym relationships provided by WordNet. This paper has also formulated the OTCM (Ontology Term Coverage Metric) to evaluate the generation of ontology at each phase.