Positioning Hotel Bintang 5 di Surabaya Menggunakan Aspect Based Sentiment Analysis dan Multiple Correspondence Analysis

Awantina, Rachma (2026) Positioning Hotel Bintang 5 di Surabaya Menggunakan Aspect Based Sentiment Analysis dan Multiple Correspondence Analysis. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003232020-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Persaingan industri perhotelan, khususnya hotel bintang 5 di Surabaya, menuntut manajemen untuk memahami persepsi pelanggan secara lebih mendalam guna menentukan strategi positioning yang tepat. Penelitian ini bertujuan untuk memetakan posisi hotel bintang 5 berdasarkan persepsi pelanggan yang terekam dalam ulasan daring melalui pendekatan text mining. Data dikumpulkan dari ulasan pelanggan di Google Maps menggunakan SerpApi, kemudian dilakukan preprocessing teks meliputi case folding, tokenizing, stopword removal, dan stemming. Identifikasi aspek layanan dilakukan dengan memanfaatkan keyword term list yang telah ditentukan sebelumnya, kemudian diperluas dengan topik tersembunyi yang diperoleh melalui metode Latent Dirichlet Allocation (LDA). Klasifikasi sentimen terhadap tiap aspek dilakukan menggunakan model Multiscale Convolutional Neural Network (MCNN) yang dilatih pada data berlabel positif dan negatif. Hasil sentimen yang diperoleh digunakan sebagai dasar untuk pemetaan posisi hotel menggunakan Multiple Correspondence Analysis (MCA). Penelitian ini menghasilkan visualisasi positioning dalam bentuk correspondence map yang memberikan gambaran komparatif persepsi pelanggan terhadap berbagai aspek layanan masing-masing hotel. Konfigurasi MCNN optimal—dengan ukuran vektor 200, ukuran filter [3, 4, 5], dan jumlah filter 100—berhasil mencapai akurasi pengujian sebesar 96,36%. Di antara kelima aspek, Comfort memperoleh performa tertinggi secara keseluruhan, sedangkan Location menunjukkan metrik yang sedikit lebih rendah. Hasil pemetaan MCA menunjukkan bahwa pada sentimen positif, aspek Location dan Cleanliness dominan pada Hotel C, sedangkan pada sentimen negatif, Hotel H cenderung diasosiasikan dengan aspek Comfort dan Service. Temuan ini diharapkan dapat menjadi dasar pengambilan keputusan strategis dalam meningkatkan daya saing industri perhotelan.
=======================================================================================================================================
Competition in the hospitality industry, particularly among five-star hotels in Surabaya, requires management to gain a deeper understanding of customer perceptions in order to formulate appropriate positioning strategies. This study aims to map the positioning of five-star hotels based on customer perceptions captured in online reviews using a text mining approach. The data were collected from customer reviews on Google Maps using SerpApi, followed by text preprocessing steps including case folding, tokenization, stopword removal, and stemming. Service aspects were identified using a predefined keyword term list and further enriched with latent topics obtained through Latent Dirichlet Allocation (LDA). Sentiment classification for each aspect was performed using a Multiscale Convolutional Neural Network (MCNN) trained on positively and negatively labeled data. The resulting sentiment outputs were then used as the basis for hotel positioning analysis through Multiple Correspondence Analysis (MCA). This study produces positioning visualizations in the form of correspondence maps, providing a comparative overview of customer perceptions across various service aspects for each hotel. The optimal MCNN configuration—with an embedding dimension of 200, filter sizes of [3, 4, 5], and 100 filters—achieved a testing accuracy of 96.36%. Among the five evaluated aspects, Comfort demonstrated the strongest overall performance, while Location exhibited slightly lower evaluation metrics. The MCA results indicate that, under positive sentiment, the Location and Cleanliness aspects are dominant for Hotel C, whereas under negative sentiment, Hotel H tends to be associated with the Comfort and Service aspects. These findings are expected to support strategic decision-making aimed at enhancing competitiveness in the hospitality industry.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	ABSA, MCA, MCNN, Positioning Hotel, Text Mining, ABSA, Hotel Positioning
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Rachma Awantina
Date Deposited:	28 Jan 2026 04:04
Last Modified:	28 Jan 2026 04:04
URI:	http://repository.its.ac.id/id/eprint/130916

Actions (login required)

View Item