SARI, MANDA (2025) Mendeteksi Komentar Bot Pada Penjualan Sebuah Produk Di Shopee Dengan Metode Gradient Boosting. S1 thesis, Universitas Malikusalleh.

	Text Cover.pdf Download (189kB)
	Text Abstrak.pdf Download (277kB)
	Text Bab I.pdf Download (389kB)
	Text Daftar Pustaka.pdf Download (404kB)
	Text Full-Text.pdf Restricted to Registered users only Download (1MB)

Abstract

Shopee merupakan salah satu platform e-commerce terbesar di Asia Tenggara yang menyediakan fitur komentar produk sebagai referensi utama bagi calon pembeli dalam menilai kualitas produk maupun reputasi penjual. Sayangnya, banyaknya komentar palsu yang dihasilkan oleh bot dengan gaya bahasa yang kaku, berulang, dan penuh pujian berlebihan menimbulkan kekhawatiran terkait keaslian ulasan yang tersedia. Hal ini dapat memengaruhi keputusan konsumen secara tidak objektif. Penelitian ini bertujuan untuk mengembangkan sistem otomatis yang mampu mendeteksi komentar bot dengan memanfaatkan algoritma Gradient Boosting. Sebanyak 3.000 komentar dikumpulkan secara manual dari berbagai kategori produk, lalu diberi label secara langsung oleh peneliti. Selanjutnya, data komentar diproses melalui tahapan pembersihan teks, tokenisasi, dan lemmatisasi agar siap dianalisis oleh model. Hasil pelatihan model menunjukkan performa yang sangat baik dengan akurasi sebesar 94,09%, presisi 95,99%, recall 83,23%, dan F1-score 89,13%. Berdasarkan hasil tersebut, dapat disimpulkan bahwa algoritma Gradient Boosting sangat efektif dalam mengklasifikasikan komentar bot dan dapat membantu meningkatkan kepercayaan serta keamanan konsumen dalam berbelanja online. Kata kunci: Komentar bot, Shopee, Gradient Boosting, Klasifikasi Teks, E-commerce Shopee is one of the largest e-commerce platforms in Southeast Asia, providing a product comment feature that serves as a primary reference for prospective buyers to assess product quality and seller reputation. Unfortunately, the prevalence of fake comments generated by bots—characterized by rigid language, repetitive patterns, and excessive praise—raises concerns about the authenticity of available reviews. This issue can negatively influence consumers’ purchasing decisions. This study aims to develop an automated system capable of detecting bot comments using the Gradient Boosting algorithm. A total of 3,000 comments were manually collected from various product categories and labeled directly by the researchers. The comment data were then processed through several stages, including text cleaning, tokenization, and lemmatization, to prepare for model analysis. The trained model demonstrated excellent performance, achieving an accuracy of 94.09%, precision of 95.99%, recall of 83.23%, and an F1-score of 89.13%. Based on these results, it can be concluded that the Gradient Boosting algorithm is highly effective in classifying bot comments and can help improve consumer trust and security in online shopping. Keywords: Bot comments, Shopee, Gradient Boosting, Text Classification, E-commerce

Item Type:	Thesis (S1)
Subjects:	H Social Sciences > HF Commerce Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software T Technology > T Technology (General)
Divisions:	Fakultas Teknik > 55201 - Jurusan Teknik Informatika
Depositing User:	Manda Sari
Date Deposited:	29 Jul 2025 06:54
Last Modified:	29 Jul 2025 06:54
URI:	https://rama.unimal.ac.id/id/eprint/13101

Actions (login required)

View Item