Ontology Learning using Hybrid Machine Learning Algorithms for Disaster Risk Management

Abstract

Disaster is inevitable but manageable thru careful planning, preparation and immediate response strategies. During typhoons, earthquakes and other calamities, agreement about language is vital to understand each other well to avoid high number of deaths, delay in access to basic needs and slow response time. However, some of the people involved in this domain find it hard to coordinate and respond to different emergency situations due to lack of familiarization and knowledge about the different terms or concepts. In disaster risk management, the consistency and reusability of the sharing of information is important to avoid possible risks. Due to this reason, an ontology is incorporated to aid in the disaster management process. The use of ontology enables quick retrieving and incorporating "consistent data" and information related to disaster management which plays an important for making decisions efficiently. This paper aims to implement and evaluate the accuracy of Support Vector Machine (SVM) and Neural Network (NN) learning-based ontology for disaster risk management to enhance the classification of concepts (keywords) generated for the domain ontology. The experiment shows that the hybrid SVM and NN machine learning algorithm outperformed the accuracy of SVM and NN based on the precision, recall and F-Measure criterion.