Sentiment Analysis Twitter Bahasa Indonesia Berbasis WORD2VEC Menggunakan Deep Convolutional Neural Network

Penulis

  • Hans Juwiantho Sekolah Tinggi Teknik Surabaya
  • Esther Irawati Setiawan Institut Teknologi Sepuluh November
  • Joan Santoso Institut Teknologi Sepuluh November
  • Mauridhi Hery Purnomo Institut Teknologi Sepuluh November

Abstrak

Media sosial sebagai media informasi dan komunikasi mulai berkembang pesat sejak internet mudah diakses. Orang dengan mudah menyatakan pendapat, ekspresi, opini, dan informasi melalui tulisan pada media sosial. Opini atau informasi pada media sosial dapat digunakan untuk menilai baik atau buruk suatu brand perusahaan. Orang cenderung jujur dalam mengungkapkan perasaan terhadap sesuatu pada media sosial. Dengan menggunakan sentiment analysis terhadap opini dari pelanggan, analisis opini dapat dilakukan secara otomatis. Perusahaan dapat secara langsung mengetahui tingkat kepuasan pelanggan dan digunakan untuk meningkatkan kualitas pelayanan hingga menaikan brand perusahaan. Penggunaan metode classical machine learning yang sudah banyak diterapkan pada sentiment analysis, tetapi metode tersebut tidak memperhatikan pentingnya urutan kata pada suatu kalimat. Metode deep learning dengan algoritme Deep Convolutional Neural Network ditawarkan untuk menjawab permasalahan tersebut dengan melakukan operasi convolution menggunakan filter sebesar ukuran window untuk mendapatkan fitur berdasarkan urutan kata. Model Word2Vec untuk Bahasa Indonesia digunakan sebagai representasi kata dalam bentuk vektor. Penggunaan Word2Vec juga mempercepat proses pelatihan dan meningkatkan akurasi algoritme Deep Convolutional Neural Network. Data yang digunakan dalam makalah ini adalah data Twitter Bahasa Indonesia dengan jumlah 999 tweet. Hasil percobaan yang telah dilakukan dengan algoritme Deep Convolutional Neural Network memiliki nilai akurasi terbaik sebesar 76,40%.

 

Abstract

Social media as information media and communication is growing rapidly since the internet is easily accessible. People easily express opinions, expressions, and information by writing on social media. Opinion or information on social media can be used to assess how good or bad a companies is. People tend to be honest in expressing feelings towards something on social media. With sentiment analysis, analysis of the opinions of customers can be done automatically. The company will know the level of customer satisfaction and can be used to improve the quality of service to raise the company's brand. The use of classical machine learning methods that have been widely applied to sentiment analysis ignoring the importance of the word order in a sentence. Deep Convolutional Neural Network algorithm is offered to answer these problems by carrying out convolution operations using filters as large as window size to get features based on word order. Word2Vec model for Indonesian is used as a word vector representation. The use of Word2Vec also reduce the training time and improve the accuracy of the Deep Convolutional Neural Network algorithm. The data used in this paper is Indonesian Twitter data with 999 tweets. The results of experiments that have been carried out with the Deep Convolutional Neural Network algorithm have the best accuracy value of 76.40%.


Downloads

Download data is not yet available.

Biografi Penulis

  • Hans Juwiantho, Sekolah Tinggi Teknik Surabaya
    Mahasiswa Pascasarjana teknologi informasi sekolah tinggi teknik surabaya

Referensi

ABIDIN, T.F., HASANUDDIN, M. dan MUTIAWANI, V., 2017. N-grams based features for Indonesian tweets classification problems. Proceedings - 2017 International Conference on Electrical Engineering and Informatics: Advancing Knowledge, Research, and Technology for Humanity, ICELTICs 2017, 2017–Octob(ICELTICs), hal.307–310.

ANASTASIA, S. dan BUDI, I., 2016. Twitter sentiment analysis of online transportation service providers. 2016 International Conference on Advanced Computer Science and Information Systems (ICACSIS), hal.359–365.

BERMINGHAM, A. dan SMEATON, A.F., 2010. Classifying Sentiment in Microblogs: Is Brevity an Advantage? Proceedings of the 19th ACM international conference on Information and knowledge management - CIKM ’10, hal.1833.

FELDMAN, R., 2013. Techniques and Applications for Sentiment Analysis. Commun. ACM, [daring] 56(4), hal.82–89. Tersedia pada: <http://doi.acm.org/10.1145/2436256.2436274>.

FININ, T., MURNANE, W., KARANDIKAR, A., KELLER, N., MARTINEAU, J. dan DREDZE, M., 2010. Annotating named entities in Twitter data with crowdsourcing. Proceedings of NAACL-HLT, [daring] 2010(January), hal.80–88. Tersedia pada: <http://dl.acm.org/citation.cfm?id=1866696.1866709>.

GO, A., BHAYANI, R. dan HUANG, L., 2009. Twitter Sentiment Classification using Distant Supervision. CS224N Project Report, 1, hal.12.

HECHT-NIELSEN, R., 1988. Theory of the backpropagation neural network. International 1989 Joint Conference on Neural Networks, hal.593–605 vol.1.

KIM, Y., 2014. Convolutional Neural Networks for Sentence Classification. CoRR, [daring] abs/1408.5. Tersedia pada: <http://arxiv.org/abs/1408.5882>.

KINGMA, D.P. dan BA, J., 2014. Adam: A Method for Stochastic Optimization. [daring] hal.1–15. Tersedia pada: <http://arxiv.org/abs/1412.6980>.

LIBRIAN, A., 2017. High quality stemmer library for Indonesian Language (bahasa), GitHub. [daring] Tersedia pada: <https://github.com/sastrawi/sastrawi> [Diakses 18 Apr 2018].

NARADHIPA, A.R. dan PURWARIANTI, A., 2012. Sentiment Classification for Indonesian Messages in Social Media. International Conference on Electrical Engineering and Informatics, (July), hal.2–5.

SANTOSO, J., SOETIONO, A.D.B., GUNAWAN, G., SETYATI, E., YUNIARNO, E.M., HARIADI, M. dan PURNOMO, M.H., 2018. Self-Training Naive Bayes Berbasis Word2Vec untuk Kategorisasi Berita Bahasa Indonesia. JNTETI, 7(2), hal.158–166.

SETIAWAN, W., 2017. Era Digital dan Tantangannya. hal.1–9.

SOMANTRI, O., APRILIANI, D., INFORMATIKA, J.T., HARAPAN, P. dan TEGAL, B., 2018. Support vector machine berbasis feature selection untuk Sentiment analysis kepuasan pelanggan terhadap pelayanan warung dan restoran kuliner kota tegal. Jurnal Teknologi Informasi dan Ilmu Komputer (JTIIK), 5(5), hal.537–548.

SRIVASTAVA, N., HINTON, G., KRIZHEVSKY, A., SUTSKEVER, I. dan SALAKHUTDINOV, R., 2014. Dropout: A Simple Way to Prevent Neural Networks from Overfitting. Journal of Machine Learning Research, [daring] 15, hal.1929–1958. Tersedia pada: <http://jmlr.org/papers/v15/srivastava14a.html>.

STATISTA.COM, 2018. Leading countries based on number of Facebook users as of October 2018 (in millions). [daring] Tersedia pada: <https://www.statista.com/statistics/268136/top-15-countries-based-on-number-of-facebook-users/> [Diakses 28 Nov 2018].

VATEEKUL, P. DAN KOOMSUBHA, T., 2016. A study of sentiment analysis using deep learning techniques on Thai Twitter data. 2016 13th International Joint Conference on Computer Science and Software Engineering (JCSSE), [daring] hal.1–6. Tersedia pada: <http://ieeexplore.ieee.org/document/7748849/>.

WINDASARI, I.P., UZZI, F.N. dan SATOTO, K.I., 2017. Sentiment Analysis on Twitter Posts : An analysis of Positive or Negative Opinion on GoJek. hal.266–269.

Diterbitkan

04-02-2020

Terbitan

Bagian

Ilmu Komputer

Cara Mengutip

Sentiment Analysis Twitter Bahasa Indonesia Berbasis WORD2VEC Menggunakan Deep Convolutional Neural Network. (2020). Jurnal Teknologi Informasi Dan Ilmu Komputer, 7(1), 181-188. https://jtiik.ub.ac.id/index.php/jtiik/article/view/1758