Analisis Sentimen Berbasis Aspek Dimensi Kualitas Dalam Meningkatkan Kualitas Produk Dan Layanan Indihome PT Telkom Indonesia Tbk Menggunakan Metode Latent Dirichlet Allocation Dan Senticircle

Hermansyah, Reza (2021) Analisis Sentimen Berbasis Aspek Dimensi Kualitas Dalam Meningkatkan Kualitas Produk Dan Layanan Indihome PT Telkom Indonesia Tbk Menggunakan Metode Latent Dirichlet Allocation Dan Senticircle. Masters thesis, Institut Teknologi Sepuluh Nopember.

	Text 09211950054005_Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (2MB) | Request a copy
	Text 09211950054005-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Persaingan kualitas produk yang begitu sengit pada industri telekomunikasi membuat setiap perusahaan berlomba-lomba menyediakan produk yang dapat diterima dan disukai oleh masyarakat tidak terkecuali pada PT Telkom Indonesia Tbk dengan salah satu produk unggulannya yakni Indihome. Dengan kualitas produk yang baik maka akan mempengaruhi kemugkinan terjadinya pembelian kembali terhadap produk tersebut, sehingga terjadilah loyalitas pelanggan terhadap produk tersebut. Upaya yang rutin dilakukan oleh Telkom untuk mengukur loyalitas pelanggannya menggunakan metode Net Promotor Score (NPS), namun metode ini masih dirasa kurang karena tidak secara spesifik menjabarkan pengaruh masing-masing aspek dimensi kualitas produk dan keterbatasan jangkauan survei. Salah satu cara untuk mengukur kualitas produk dengan jangkauan yang luas dengan menggunakan metode analisis sentimen pada media sosial. Analisis sentimen banyak digunakan karena kemampuannya untuk mendapatkan wawasan yang kaya tentang detail dan alasan tren pasar yang tidak jelas, hal tersebut menjadi penting bagi pihak bisnis dalam meningkatkan kualitas produknya dan layanannya dipasaran.
Penelitian ini melakukan analisis sentimen pada produk Indihome PT. Telkom Indonesia Tbk berdasarkan ulasan yang ditulis dalam bentuk tweet pada Twitter. Setiap tweet mengenai Indihome dikumpulkan menjadi data lalu di kategorisasikan bedasarkan delapan aspek dimensi kualitas (Performance, Reliability, Features, Perceived Quality, Serviceability, Durability, Conformance dan Aesthetics) dengan beberapa tahapan, pertama keyword term aspek dimensi kualitas produk diperluas dengan metode Term Frequency Inverse Cluster Frequency (TF-ICF). Kemudian, Lantent Dirichelet Allocation (LDA) untuk menemukan hidden topic dari tweet dan mengitung Sematic Similarity untuk mengkategorikan setiap aspeknya. Selanjutnya diujikan perbandingan dari kombinasi jenis algoritma klasifikasi : SentiWordNet, SVM dan Random Forest dengan menggunakan faktor evaluasi sentimen. Hasil masing-masing model klasifikasi selanjutnya akan ditentukan kembali polaritasnya menggunakan SentiCircle dengan harapan meningkatkan akurasinya. Perhitungan Net Brand Reputation (NBR) akan dibandingkan degan Net Promotor Score (NPS) setelah hasil akhir sentimen berbasis aspek diperoleh.
Hasil penelitian menunjukan bahwa kategorisasi aspek dengan pengabungan model LDA+Semantic Similarity dan digabungan denngan TF-ICF 100% untuk perluasan term memberikan performasi sebesar 70.1%. Kemudian, pada klasifikasi sentimen menunjukan bahwa gabungan model SentiWordNet+SVM dibantu dengan SentiCircle menghasilkan performa sebesar 96.3%. Selanjutnya, kedua model tersebut digunakan untuk menganalisis sentimen berdasarkan delapan aspek dimensi kualitas. Peneliti mendapatkan bahwa aspek dimensi kemudahan layanan (Serviceability) memiliki rata-rata sentimen negatif yang mencapai 20.82% dibandingkan dengan aspek dimensi lainnya (Performance : 12.86%, Reliability : 2.53%, Features : 13.28%, Perceived Quality : 0.99%, Durability 4.44%, Conformance : 2.29% dan Aesthetics : 5.86%). Hasil yang sama juga diperoleh pada perhitungan NBR, sentimen negatif mendominasi dengan aspek dimensi kinerja (Performance) yang terburuk -11.46. Berdasarkan penelitian ini dapat disimpulkan bahwa pihak manajemen PT. Telkom Indonesia Tbk perlu melakukan analisis sentimen secara rutin yang bertujuan untuk pengukuran kualitas produk di media sosial dan menjadi pelengkap metode NPS. Hasil penelitian juga menunjukkan bahwa manajemen Telkom perlu mengaudit dan mengevaluasi kembali poeple, proses and technology untuk dapat meningkatkan kualitas dan image produk Indihome.
====================================================================================================
The fierce competition for product quality in the telecommunications industry makes every company compete to provide products that are acceptable and liked by the public, including PT Telkom Indonesia Tbk with one of its superior products, Indihome. With good product quality, it will affect the possibility of repurchasing the product, so that there is customer loyalty to the product. Efforts are routinely made by Telkom to measure customer loyalty using the Net Promoter Score (NPS) method, but this method is still lacking because it does not specifically describe the influence of each aspect of product quality dimensions and the limitations of survey coverage. One way to measure product quality with a wide range is using the sentiment analysis method on social media. Sentiment analysis is widely used because of its ability to gain rich insight into the details and reasons for unclear market trends, it becomes important for businesses in an effort to improve the quality of their products and services in the market.
This study conducted a sentiment analysis on the product Indihome PT. Telkom Indonesia Tbk based on a review written in the form of a tweet on Twitter. Each tweet about Indihome is collected into data and then categorized based on eight aspects of the quality dimension (Performance, Reliability, Features, Perceived Quality, Serviceability, Durability, Conformance and Aesthetics) in several stages, firstly, the keyword term aspects of product quality dimensions are expanded with the Term Frequency Inverse method. Cluster Frequency (TF-ICF). Then, Lantent Dirichelet Allocation (LDA) is used to find hidden topics from tweets and calculates Sematic Similarity to categorize each aspect. Furthermore, the comparison of the combination of classification algorithm types: SentiWordNet, SVM and Random Forest is tested by using a sentiment evaluation factor. The results of each classification model will then be re-determined using the SentiCircle polarity in the hope of increasing its accuracy. The calculation of the Net Brand Reputation (NBR) will be compared with the Net Promoter Score (NPS) after the final result of aspect-based sentiment is obtained.
The results showed that the categorization of aspects by merging the LDA+Semantic Similarity model and combining it with TF-ICF 100% for term expansion gave a performance of 70.1%. Then, the sentiment classification shows that the combined SentiWordNet+SVM model assisted by SentiCircle produces a performance of 96.3%. Furthermore, the two models are used to analyze sentiment based on eight aspects of the quality dimension. The researcher found that the serviceability dimension has an average negative sentiment that reaches 20.82% compared to other dimensions (Performance: 12.86%, Reliability: 2.53%, Features: 13.28%, Perceived Quality: 0.99%, Durability 4.44% , Conformance : 2.29% and Aesthetics : 5.86%). The same result is also obtained in the calculation of NBR, negative sentiment dominates with the worst aspect of the performance dimension -11.46. Based on this research, it can be concluded that the management of PT. Telkom Indonesia Tbk needs to conduct regular sentiment analysis aimed at measuring product quality on social media and as a complement to the NPS method. The results also show that Telkom management needs to audit and re-evaluate people, processes and technology to improve the quality and image of Indihome products.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Kualitas Produk, Delapan Aspek Dimensi Kualitas, Analisis Sentimen, LDA, SentiCircle, NBR. Product Quality, Eight Aspects of Quality Dimensions, Sentiment Analysis, LDA, SentiCircle, NBR
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD9980.5 Service industries--Quality control. T Technology > T Technology (General) > T58.6 Management information systems
Divisions:	Faculty of Creative Design and Digital Business (CREABIZ) > Technology Management > 61101-(S2) Master Thesis
Depositing User:	Reza Hermansyah
Date Deposited:	06 Aug 2021 02:44
Last Modified:	06 Aug 2021 02:44
URI:	http://repository.its.ac.id/id/eprint/84973

Actions (login required)

View Item