Sentimen Analisis Pada Ulasan Google Play Store Menggunakan Metode Naïve Bayes
DOI:
https://doi.org/10.34818/eoe.v7i2.13134Abstract
Abstrak Ulasan atau biasa disebut review merupakan salah satu fitur penting dari aplikasi yang ada pada Google Play Store. Fitur ini dapat digunakan oleh para pengguna untuk menilai serta memberikan pendapat berupa teks ulasan terhadap aplikasi yang digunakan. Namun untuk beberapa kasus, terdapat teks ulasan yang tidak sesuai dengan nilai atau rating yang diberikan. Contohnya jika pengguna memberikan rating bintang 5 namun memberikan teks ulasan yang bernada negatif. Penelitian ini membangun mesin klasifikasi yang dikhususkan untuk membandingkan teks ulasan yang diberikan oleh pengguna dengan rating yang diberikan. Metode yang digunakan yaitu Naïve Bayes, khususnya Multinomial Naïve Bayes untuk memudahkan proses klasifikasi karena metode ini dikhususkan untuk klasifikasi teks. Dari hasil penelitian, didapat akurasi setinggi 78,9%, untuk ulasan yang memiliki rating bintang 5 dan bintang 1. Tetapi akurasi menurun mencapai 73,7% untuk semua rating ulasan terkecuali bintang 3. Kata kunci : klasifikasi, sentimen analisis, ulasan aplikasi, naïve bayes, multinomial Naïve Bayes Abstract Review is one of the most important feature of application on Google Play Store. Review can be used by user to rate and share their experince with the application with text review. But for some cases, there are some review that don’t align with the star rating that was given from user. For example, if user gave a 5 star rating but their text review are written as negative experience. This research builds a classifier that can be used to compare the text review with the star rating. The method that used in this research is Naïve Bayes, especially Multinomial Naïve Bayes to ease the classification process because Multinomial Naïve Bayes is specialized in text classification. From this research, the highest accuracy that produced is 78,9% for review with 5 star and 1 star review, and the lowest with 73,7% for all rating except 3 star review. Keywords: classification, sentiment analysis, application review, naïve bayes, multinomial naïve bayes
Downloads
References
[1] Zendesk; Dimensional Research, “What is the Impact of Customer Service on Lifetime Customer Value?,” Zendesk, [Online]. https://www.zendesk.com/resources/customer-service-and-lifetimecustomer-value
[2] Liu, B., Sentiment analysis and Subjectivity. Handbook of natural language processing, 2, pp.627-666, 2010.
[3]B. Pang, L. Lee dan S. Vaithyanathan, “Thumbs up? Sentiment Classification using Machine Learning Techniques,” dalam Empirical Methods on Natural Language Processing, New York, 2002
[4] Wahid, D. H., & Azhari, S. N. (2016). Peringkasan Sentimen Esktraktif di Twitter Menggunakan Hybrid TFIDF dan Cosine Similarity. IJCCS (Indonesian Journal of Computing and Cybernetics Systems), 10(2), 207-218.
[5] J.Song, K.T. Kim, B. Lee, S. Kim, dan H.Y. Youn, “A novel classification approach based on Naïve Bayes for Twitter sentiment analysis” dalam Ksii Transactions On Internet And Information Systems Vol. 11, No. 6, 2017.
[6]F.I. Faizal, T.Mohamad, D.H. Anggit, Analisis Sentimen Review Aplikasi Ruangguru Menggunakan Algoritma Support Vector Machine, Jurnal Bisnis, Manajemen Informatika Vol. 11, No.3, 2020
[7] Feldman, R & Sanger, J. The Text Mining Handbook : Advanced Approaches in Analyzing Unstructured Data. Cambridge University Press, New York, 2007.
[8] D. Kelly et al. Methods for evaluating interactive information retrieval systems with users. Foundations and Trends R in Information Retrieval, 3(1–2):1–224, 2009.
[9] J. M. Zhishuo Liu, Qianhui Shen. Extracting implicit features based on association rules. 2018.
[10] Liu, B. 2012. Sentiment analysis and opinion mining. Synthesis Lectures on Human Language Technologies



