PENERJEMAH BAHASA BESEMAH BERBASIS MACHINE LEARNING DENGAN ALGORITMA MODEL ENCODER-DECODER

Abstract

Regional languages are one of the cultural elements in Indonesia, reflecting the diverse traditions of various regions across the archipelago. The Besemah language, spoken by the Besemah ethnic group, is used daily by the residents of Pagar Alam City and Lahat Regency. Although there are differences in pronunciation and vocabulary between villages, the Besemah language still shares a common linguistic root. The residents of Pagar Alam, particularly those in the mountainous areas of South Sumatra Province, are among the native speakers of the Besemah language. To enhance the preservation of the Besemah language, this study utilizes a translation system between Besemah and Indonesian. The development of this translation system, in line with technological advancements, will facilitate users in understanding translations between the Besemah language and Indonesian. The researchers chose the Encoder-Decoder algorithm because this architecture has been proven effective in various translation tasks and can produce more accurate translations compared to existing methods such as the Edit Distance algorithm and Convolutional Neural Networks (CNN). This algorithm consists of two main components: the Encoder, which processes the input document and converts it into a translation for each word in the Besemah language, and the Decoder, which takes the data representation generated by the Encoder and processes it into a translation output in Indonesian and vice versa.