Penentuan Algoritma Similarity Yang Akurat Pada Sistem Berbasis Cased Based Reasoning Untuk Identifikasi Ikan

Abstract

Indonesia as an archipelagic country has enormous and diverse fishery potential, Indonesia has 17,508 islands with a coastline of 81,000 km and 70% (5.8 million km2) of the area of Indonesia are oceans, as for the diversity of marine resources for various types of fish. distinctive features to recognize and understand a species in detail. The identification of the types of fish that are computational is still limited. In this research, the analysis used is the Case Based Reasoning (CBR) system. CBR is a case-based reasoning that has a knowledge- based problem solving method to study and solve problems based on past experiences. CBR is a computational model to imitate human reasoning to make it easier to find cases based on similarities, then the case based reasoner looks for cases that exist on a case basis to find cases that are similar to the problem at hand (retrieve). Therefore, the CBR process is often referred to as "4 Re" namely Retrive, Retain, Revise, Reuse. In the paradigm of problem solving, a new problem is solved by comparing it with cases in the past and reusing existing cases to solve a problem now. The purpose of this study is to evaluate and compare the Weighted Euclidean Distance (WED) Algorithm, the Hamming and Levenshtein Distances (HLD) Algorithm, the Cosine Coefficient Algorithm for Text-Based Cases (CCFTBC) and the k- Nearest Neighbor (k-NN) Algorithm for identification of fish species. The result of this research is to find a fast and accurate selection of algorithm comparison results in CBR systems.