Ecology, Environment and Conservation Paper

IMPROVING THE CAPABILITY OF TESTING OF EXPORT FISHERY PRODUCTS WITH INTELLIGENT COMPUTER SYSTEMS

Sebastianus Adi Santoso Mola

Abstract

Fishery products to be exported must pass testing at the Fish Quarantine and Quality Control Agency to ensure that there is no spread of disease to the export destination area. The spread of the disease to the destination area can cause an outbreak, destroying the environment and the economy of the area. The testing process which takes time can result in increased costs and risk of death for live fishery products. In this research, a case-based smart system was developed to reduce the testing process time. The intelligent system is constantly learning all the time to improve the ability to identify disease, both in terms of computing speed and in reducing human error. This study aims to obtain good indexing techniques based on fish species. The results of the indexing test show excellent classification capabilities for NBC, J48 and MLP for balanced data sets with an average TP above 90%. The NBC method is chosen for the indexing stage by considering its high classification capability, classification capability with limited data and very fast classification time. The case retrieval results show consistency that NBC is superior in the similarity of new cases and those in the case base.