Rekomendasi Produk Berdasarkan Preferensi Konsumen dan Analisis Sentimen Berbasis Aspek pada Ulasan Pembelian

Nabila, Ifta Jihan (2021) Rekomendasi Produk Berdasarkan Preferensi Konsumen dan Analisis Sentimen Berbasis Aspek pada Ulasan Pembelian. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 05111740000034-Undergraduate_Thesis.pdf]
Preview
Text
05111740000034-Undergraduate_Thesis.pdf - Accepted Version

Download (4MB) | Preview

Abstract

Marketplace di Indonesia selalu mengalami perkembangan dari tahun ke tahun. Beberapa marketplace seperti Shopee dan Bukalapak merupakan marketplace yang memiliki paling banyak pengunjung pada tahun 2020. Banyaknya pilihan produk yang beredar di marketplace mengakibatkan konsumen akan membandingkan produk dari toko lain di marketplace yang sama dan juga dari marketplace lain.Sistem yang dibangun dapat memberikan rekomendasi produk yang sesuai dengan jenis produk yang ingin dicari atau preferensi konsumen dari kumpulan produk yang ada di Shopee dan Bukalapak. Rekomendasi produk diperoleh dari kumpulan produk yang relevan dengan preferensi serta mempertimbangkan harga, rating, dan ulasan produk dari setiap toko pada kedua marketplace tersebut. Ulasan produk kemudian dianalisa lebih lanjut menggunakan model aspect-based sentiment analysis (ABSA).Kategori aspek yang digunakan untuk analisa ulasan produk ini yaitu kesesuaian deskripsi produk, kesesuaian harga produk, kualitas produk, pengiriman, dan pelayanan. Data ulasan produk ini diolah lebih lanjut untuk menangani permasalahan data tak berlabel dengan metode semi-supervised learning serta penerapan metode augmentasi pada data tidak seimbang. Pembuatan model ABSA ini dilakukan menggunakan beberapa fitur ekstraktor dan metode klasifikasi multi-label dengan pendekatan deep learning. Hasil uji coba skenario terbaik dengan menggunakan FastText dan Bi-GRU menghasilkan average micro f1-score dan average macro f1-score sebesar 92.10% dan 92.92%.
====================================================================================================
Marketplaces in Indonesia have always been developing from year to year. Several marketplaces such as Shopee and Bukalapak are marketplaces with the most visitors in 2020. There are many product choices in the marketplace, resulting in consumers comparing products from other stores in the same marketplace and also from other marketplaces.
The system built can provide product recommendations according to the type of product that consumers want to find or consumer preferences from a collection of products in Shopee and Bukalapak. Product recommendations are obtained from a collection of products that are relevant to preferences and consider prices, ratings, and product reviews from each store in the two marketplaces. The product reviews will then be analyzed further using an aspect-based sentiment analysis (ABSA) model.
The aspect categories used for this product review analysis are product description relevance, product price relevance, product quality, delivery, and service. This product reviews data will be further processed using the semi-supervised learning method to overcome the problem of unlabeled data and the application of the augmentation method to overcome imbalanced data. The making of this ABSA model will be carried out using a multi-label classification method with a deep learning approach. The best scenario result was using FastText and Bi-GRU. The scores obtained were the average micro f1 score and the macro f1 average score of 92.10% and 92.92%.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: aspect-based sentiment analysis, semi-supervised learning, text data augmentation, multi-label classification, FastText, Bidirectional GRU, aspect-based sentiment analysis, semi-supervised learning, augmentasi data teks, klasifikasi multi-label, FastText, Bidirectional GRU
Subjects: H Social Sciences > HF Commerce > HF5415.32 Consumers' preferences
Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science)
R Medicine > R Medicine (General) > R858 Deep Learning
T Technology > T Technology (General) > T57.5 Data Processing
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User: IFTA JIHAN NABILA
Date Deposited: 05 Aug 2021 01:06
Last Modified: 10 Sep 2024 01:18
URI: http://repository.its.ac.id/id/eprint/84881

Actions (login required)

View Item View Item