Supervised Learning for Demospongiae Identification using Graph Mining Technique

Abstract

This research purposes a graph mining technique to identify the particular characteristic of Demospongiae by supervised learning method. The sponge dataset contained 7 families belonging to the Demospongiae class was collected from the Mediterranean and Atlantic oceans. The dataset of Demospongiae was performed on graph dataset and then was analyzed to identify the particular characteristic by using graph based supervised learning method. The learned substructures can identify the unique of a specific feature for each family and can use to develop the prototype of knowledge-based expert system for Demospongiae identification. The prototype was evaluated by measuring the efficiency from the ability to classify the sponge family. The experimental results showed the identification performance accuracy is 76.47%. This indicated that graph mining technique by supervised learning method is valid and practicable for Demospongiae identification.