Komering - Indonesian Machine Translation Using Embedding RNN

The Unitary State of the Republic of Indonesian has diverse language cultures across its regions. One of these is the Komering language culture in South Sumatra. Although the Komering tribe continues to use the Komering language, its speakers have decreased over time. Hence, developing a machine translation model for the Komering language is necessary. This research aims to develop a neural machine translator using the embedding RNN algorithm. The study evaluates several embedding RNN models to identify the model with the best accuracy performance. The best model produces results with an accuracy reaching 0.9583 during the training stage and a validation accuracy of 0.8866. These values indicate that the model can effectively translate sentences from Indonesian to Komering and vice versa.