Oktavia, Rani (2017) Penilaian Kualitas Ulasan Pelanggan Berdasarkan Karakteristik Struktural, Metadata, Dan Keterbacaan. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
5213100122_Undergraduate_Theses.pdf - Published Version Download (2MB) | Preview |
Abstract
Ulasan online pelanggan memiliki peran penting dalam proses keputusan pembelian produk. Sekarang ini, semakin banyak ulasan online yang tersedia di pasar online sehingga pelanggan dapat membacanya untuk lebih memahami dengan baik mengenai produk atau jasa yang akan dibeli. Ulasan pelanggan menjadi informasi tambahan yang penting disamping informasi yang disediakan pasar online seperti deskripsi produk, ulasan dari ahli, dan rekomendasi dari sistem. Namun, karena semakin banyak ulasan pelanggan yang tersedia sekarang ini, muncullah pertanyaan apakah setiap ulasan tersebut berkualitas dan berguna bagi pelanggan lain. Berdasarkan hasil survei, sebanyak 87% pembeli akan membaca paling banyak 10 ulasan untuk menentukan keputusan pembelian. Hal ini dapat menimbulkan permasalahan karena adanya kemungkinan ulasan pelanggan yang baru ditulis dan berkualitas, namun belum populer sehingga tidak terbaca oleh pelanggan lain.
Oleh karena itu, diperlukan penilaian kualitas konten ulasan berdasarkan tiga karakteristik yaitu struktural, meta-data, dan keterbacaan menggunakan weighted sum yang dapat mengevaluasi beberapa alternatif berdasarkan kriteria tertentu. Kualitas konten ulasan juga akan dinilai menggunakan metode Support Vector Machine. Metode ini digunakan karena data memiliki fitur yang banyak yang mencakup kategori struktural, meta-data, dan keterbacaan sehingga metode ini dapat digunakan untuk mengklasifikasi kualitas ulasan pelanggan. Hasil yang didapatkan menunjukkan bahwa perhitungan dengan weighted sum, diperoleh nilai kualitas ulasan tertinggi sebesar 0,736 dari skala 1 dan nilai kualitas ulasan terendah sebesar -0.104. Kategori yang paling mempengaruhi penilaian kualitas ulasan adalah nilai keterbacaan automated readability index, sedangkan kategori kegunaan tidak begitu mempengaruhi penilaian. Sedangkan berdasarkan hasil klasifikasi menggunakan support vector machine untuk memprediksi kegunaan suatu ulasan, diperoleh nilai keakuratan paling tinggi menggunakan kernel polynomial dengan nilai sebesar 94.4773% .
===============================================================================================
Customer online review has the important role in product decisions process. Nowadays, there are a lot of customer online reviews available in online marketplace so that customer can read those reviews to better understand about product or service that will they purchase. Customer online reviews become important additional information besides product description, expert reviews, and recommendation from systems. However, there are a lot of reviews that become questions are these reviews have good quality and useful for other customers. Based on a survey, 87% customers will read at least 10 reviews before deciding to buy a product. This thing can become problematic because there's a possibility for new customer online reviews that have good quality but hasn't popular yet so this reviews might be miss read by other customers.
Therefore, it is necessary to evaluate the quality of content review based on three characteristics which are structural, metadata, and readability using weighted sum that can evaluate some alternatives based on certain criteria. The quality of content review will also be evaluated using Support Vector Machine. This method is used because the data has a lot of features including structural category, metadata, and readability so this method can be used for classifying the quality of online reviews. The results show that the calculation
of the weighted sum, obtained the highest quality review value of 0.736 from scale 1 and the lowest quality review value of -0.104. The category that most affects the quality rating of the review is the value of automated readability index readability, whereas the usability category does not significantly affect the assessment. While based on the classification results using the support vector machine to predict the usefulness of a review, obtained the highest accuracy value using polynomial kernel with a value of 94.4773%.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSSI 025.3 Okt p-1 |
| Uncontrolled Keywords: | Ulasan Online Pelanggan, Kualitas Ulasan, Weighted Sum, Support Vector Machine |
| Subjects: | T Technology > T Technology (General) T Technology > TS Manufactures > TS156 Quality Control. QFD. Taguchi methods (Quality control) |
| Divisions: | Faculty of Information Technology > Information System > 57201-(S1) Undergraduate Thesis |
| Depositing User: | Oktavia Rani |
| Date Deposited: | 31 Oct 2017 04:40 |
| Last Modified: | 05 Mar 2019 03:31 |
| URI: | http://repository.its.ac.id/id/eprint/42412 |
Actions (login required)
 |
View Item |