Ayu Purwarianti
Automatic question generation is defined as the task of automating the creation of question given a various of textual data. Research in automatic question generator (AQG) has been conducted for more than 10 years, mainly focused on factoid question. In all these studies, the state-of-the-art is attained using sequence-to-sequence approach. However, AQG system for Indonesian has not ever been researched intensely. In this work we construct an Indonesian automatic question generator, adapting the architecture from some previous works. Most of the previous studies stated that models developed with deep learning approach heavily outperform their rule-based counterpart, with higher flexibility and larger domain scope too. Acknowledging this phenomenon, we propose an AQG using sequence to sequence approach which mainly focused to create an Indonesian AQG. Generally, we design two main models, RNN-based, and Transformer-based. We use paragraph context sentences along with some additional linguistic features: answer location (ans), case indicator (case), part of speech (POS), and named entity (NE) as the input, and question sentences as the target. In summary we use sequence-to-sequence approach using BiGRU, BiLSTM, and Transformer with additional linguistic features, copy mechanism, and coverage mechanism. Since there is no public large dan popular Indonesian dataset for question generation, we use translated SQuAD v2.0 factoid question answering dataset, with additional Indonesian TyDiQA dev set for testing. Our system achieves BLEU1, BLEU2, BLEU3, BLEU4, and ROUGE-L score at 38,35, 20,96, 10,68, 5,78, and 43,4 for SQuAD, and 39.9, 20.78, 10.26, 6.31, 44.13 for TyDiQA, respectively. Our system is found to perform well when the expected answers are a named entity and are syntactically close with the context explaining them.
Penerapan Karya Tulis
AQG system for Indonesian has not ever been researched intensely.