Utilizing Morphological Features for Part-of-Speech Tagging of Bahasa Indonesia in Bidirectional LSTM

Research in the area of Part of Speech (PoS) Tagging has been widely explored especially for high resource language, such as English. However, there are only a small number of studies that have been conducted for Bahasa Indonesia. In this study, we present our experiment on utilizing morphological features for PoS tagging of Bahasa Indonesia in Bidirectional Long Short Term Memory architecture. Three different features including prefix, suffix, and capitalization have been examined. The results of our study show that combining morphological features with word embedding is effective for improving the tagger performance. Our study also provides more detailed explanation on which morphological features are useful for the PoS tagging task.