Pengenalan Intent pada Natural Language Understanding Berbahasa Indonesia dengan Menggunakan Metode Convolutional Neural Network

Daniel Adi, Leo Willyanto Santoso, Alvin Nathaniel Tjondrowiguno

Abstract


To keep up with technological developments and people behavior, intelligent bot has become part of the business world which help them maintain good relation with their customer. Unfortunately, resource for intelligent bot in Indonesian language is very scarce compared to High Resource Language like English. Therefore further research about Natural Language Understanding in Indonesian language is needed. We use Convolutional Neural Network method to train our model. Model consist of embedding layer, convolutional layer, max pooling, flatten, dropout, and softmax layer. In the process of making model, there are many variable that can be tested such as dropout, number of filter, size of filter, etc. This research show that the amount and quality of data for each category can affect how a model understand the feature of each category which affect the overall precision. The quality of word2vec, one of the most important resource in the model can give significant impact on precision. The size of dropout can affect how the model understand the important feature of data. From various tests, we found that the best precision is 93 %.

 


Keywords


Artificial Neural Network; Keras; Convolutional Neural Network; Intelligent Bot

Full Text:

PDF

References


Barkin, J. 2016. When Bots Fail At Conversation. URL= https://medium.com/@joshbarkin/when-bots-fail-at-conversation-d7419605f5cc

Bashira, A. M., Hassan, A., Rosman, B., Duma, D., & Ahmed, M. (2018). Implementation of A Neural Natural Language Understanding. ACLing 2018, (p. 228). Dubai.

Eka, R. 2017. Pengembangan Layanan Berbasis Chatbot di Indonesia Sebagai “The Next Big Thing”. URL= https://dailysocial.id/post/pengembangan-layanan-berbasis-chatbot-di-indonesia-sebagai-the-next-big-thing

Hirschberg, J., & Manning, C. D. 2015. Advances in natural language processing. URL= https://cs224d.stanford.edu/papers/advances.pdf

Kim, Y. 2014. Convolutional Neural Networks for Sentence Classification. 4. arXiv:1408.5882

Oracle. 2016. Can Virtual Experience Replace Reality ? URL= https://www.oracle.com/webfolder/s/delivery_production/docs/FY16h1/doc35/CXResearchVirtualExperiences.pdf

Zhang, Y., & Wallace, B. C. 2016. A Sensitivity Analysis of (and Practitioners’ Guide to) Convolutional Neural Network for Sentence Classification. 6. arXiv:1510.03820


Refbacks

  • There are currently no refbacks.


Jurnal telah terindeks oleh :