Mobile expert systems for bamboo identification using rule based reasoning

Abstract

Indonesia is tropical country which has high plant diversity including bamboo. This condition make difficult to identification process. Bamboo species in Indonesia is estimated at about 159 species out of the total 1250 species around the world. The bamboo identification method usually takes a long time, relatively expensive, complicated, and requires information from bamboo expert. By transferring the knowledge of bamboo experts in recognizing the characteristics of bamboo such as roots, shoots, stems, branches, and leaves, then identification can be done without having to bring an expert directly. Data in mobile expert system for the bamboo identification is divided into 6 morphologies, 31 feature attributes, and 188 characteristic. This system is built with rule based reasoning method that modified with greedy algorithm to compare the weight difference of bamboo species. The experiments were performed with 143 bamboo test cases have shown 100% high accuracy rate for the same system input 100%. Furthermore, the accuracy rate reaches 100% for 50% input data equal to the knowledge base and only 50% answer filled, so the user does not need to answer the whole question to get accurate identification result. This result shows that the mobile expert system for bamboo identification with rule based reasoning generated results more reliable than human expert.