ModifiedECS (mECS) Algorithm for Madurese-Indonesian Rule-Based Machine Translation
Fika Hastarita Rachman, Noor Ifada, Sri Intan Wahyuni, Gita Dharma Ramadani, Adrian Pawitra · 2022
The Madurese language is a local Indonesian culture that needs to be preserved. The morphology of Madurese words is unique, and there are several forms of words: affixation, root word, degree modifier, and reduplication word. Each word has a pattern that Rule-Based Machine Translation (RBMT) uses to ensure accurate translation. RBMT requires a stemming process to convert each word into its root word. The Madurese language’s morphology has several unique characteristics that need careful study, including the affix. An affix attached to a word can have different meanings depending on the conditions. This paper develops a new stemming algorithm called modified ECS (mECS), i.e., a modification of ECS stemming combined with the concept of Rule-Based word morphology. The system was tested using 50 Madurese language sentences randomly selected from a Madurese language textbook for the 5th-grade elementary school Madura language learning textbook. The accuracy of proposed RBMT system that implements the newly developed mECS stemming algorithm is 85%.