Word segmentation, part of speech of tagging and dependency parsing are the important role in Natural Language Processing (NLP). The POS and dependency parsing information are also necessary for NLP’s applications such as machine translation (MT), information retrieval (IR), etc. Although there are many research efforts in this process, there is still necessary to develop standard model for the Myanmar Language. This system uses Recurrent Neural Network (RNN) to keep away from errors and improve segmentation by utilizing POS data. In this paper, this system compares BILSTM and Hidden Markov Model (HMM) and the performances showed that precision, recall, F1 score, support and confusion matrix by using MLPOS.