Named-Entity Recognition for Indonesian Language using Bidirectional LSTM-CNNs

Abstract

In this paper, we describe the implementation of Named-Entity Recognition (NER) for Indonesian Language by using various deep learning approaches, yet mainly focused on hybrid bidirectional LSTM (BLSTM) and convolutional neural network (CNN) architecture. There are already several developed NERs dedicated to specific languages such as English, Vietnamese, German, Hindi and many others. However, our research focuses on Indonesian language. Our Indonesian NER is managed to extract the information from articles into 4 different classes; they are Person, Organization, Location, and Event. We provide comprehensive comparison among all experiments by using deep learning approaches. Some discussions related to the results are presented at the end of this paper. Through several conducted experiments, Indonesian NER has successfully achieved a good performance.