Open AccessThe Combination of YAKE and Language Processing for Unsupervised Term Extraction Ontology Learning
Author(s) -
Rajif Agung Yunmar,
Andika Setiawan,
Hartanto Tantriawan
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/537/1/012023
Subject(s) - computer science , correctness , term (time) , ontology , natural language processing , artificial intelligence , ontology learning , information extraction , the internet , information retrieval , machine learning , process ontology , suggested upper merged ontology , world wide web , domain knowledge , algorithm , philosophy , physics , epistemology , quantum mechanics
Information that is spread on the internet is available in the form of unstructured texts that can only be understood by humans, but difficult for machines to understand. Ontology learning is a method that can transform information in unstructured forms, into information that can be understood by machines, namely ontology. In ontology learning, the extraction term is one of the stages that must be passed. This stage produces important terms related to a topic before finally being grouped in certain classes. In this study, the term extraction method used is YAKE. The contribution of this research is to investigate the effects of language processing such as stemming and stopword removal when combined with the YAKE method at the term extraction stage. The language processing technique is then applied to the corpus of the test, after that it is as the input to the YAKE term extraction. Testing is conducted with several scenarios, namely: plain YAKE, stemming+YAKE, stopword removal+YAKE, or a combination three of them. These extraction scenario are evaluated by expert for measure the term correctness. The research shows that the combination of stopword removal+YAKE provide the best accuracy of 48%.