Analysis Ontology-Based Simple Closed Domains Question-Answer Application
Ade Hodijah, Urip Teguh Setijohatmo, Jonner Hutahaean, Ade Chandra Nugraha, Naufal Dalhats Zhalifunnas · Advances in engineering research/Advances in Engineering Research · 2024
Time performance and precision became crucial factors in questionanswer applications.This research compares two ontology-based models of the simple question-answer (QA) application.The domain of the application is movie information.The first model used lightweight natural language processing (NLP) and a detail movie ontology combination (shallow NLP movie ontology).The other designed lightweight movie ontology was suitable with the IndoBERT mechanism (IndoBERT movie ontology).The research used a run-time experiment to get the best performance model using the IMDB dataset.The query processing experiment controlled the amount of data, while the precision experiment was based on the output generated by the Shallow NLP movie ontology and the IndoBERT movie ontology.The results show that the Shallow NLP movie ontology was better for the execution times and more precision (as long as supported by valid ontology instances for all of the questions) than the IndoBERT movie ontology.In contrast, the precision of IndoBERT movie ontology was more powerful NLP than the Shallow NLP movie ontology.This research is expected to provide an empirical contribution regarding building and developing a simple QA application and ontology for closed domains.