Sistem Rekomendasi Produk Atasan Wanita Menggunakan Aspect-Based Sentiment Analysis (Absa) Pada E-Commerce

Nashirah, Dyas Amorita Radhwa (2025) Sistem Rekomendasi Produk Atasan Wanita Menggunakan Aspect-Based Sentiment Analysis (Absa) Pada E-Commerce. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027211009-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Pesatnya pertumbuhan e-commerce di Indonesia mempersulit konsumen dalam menentukan produk yang layak dibeli. Untuk mengatasi hal tersebut, dikembangkan sistem analisis sentimen real-time berbasis Aspect-Based Sentiment Analysis (ABSA) yang mengekstraksi opini konsumen berdasarkan aspek produk. Sistem ini difokuskan pada ulasan produk pakaian dengan tiga aspek utama, yaitu bahan, jahitan, dan kesesuaian produk. Model IndoBERTweet digunakan untuk analisis sentimen aspek, dengan capaian F1-score sebesar 0,88, menunjukkan performa yang baik dalam memahami ulasan dengan bahasa informal. Selanjutnya, metode Simple Additive Weighting (SAW) digunakan untuk menghitung skor rekomendasi produk berdasarkan kombinasi bobot aspek dan selisih ulasan positif-negatif yang telah dinormalisasi. Hasil implementasi menunjukkan sistem mampu melakukan crawling data, analisis sentimen, serta penyajian rekomendasi produk secara efektif, sehingga dapat membantu konsumen dalam mengambil keputusan pembelian yang lebih tepat di platform e-commerce.
=================================================================================================================================
The rapid growth of e-commerce in Indonesia has made it increasingly difficult for consumers to determine which products are worth purchasing. To address this issue, a real-time sentiment analysis system based on Aspect-Based Sentiment Analysis (ABSA) was developed to extract user opinions from product reviews based on specific aspects. This system focuses on clothing product reviews by analyzing three key aspects: material quality, stitching, and product conformity to descriptions or images. The sentiment analysis leverages the IndoBERTweet model, which demonstrates strong performance in handling informal language reviews, achieving an F1-score of 0.88. Furthermore, the Simple Additive Weighting (SAW) method is applied to calculate product recommendation scores by combining aspect weights with normalized sentiment differences between positive and negative reviews. The results show that the system effectively performs data crawling, sentiment analysis, and recommendation presentation, supporting consumers in making more accurate purchasing decisions on e-commerce platforms.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	IndoBERT, Crawling, Aspect based Sentiment Analysis, website.
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence T Technology > T Technology (General) > T59.7 Human-machine systems.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Dyas Amorita Radhwa Nashirah
Date Deposited:	30 Jul 2025 02:48
Last Modified:	30 Jul 2025 02:48
URI:	http://repository.its.ac.id/id/eprint/122882

Actions (login required)

View Item