Pengelompokan Produk Skincare Berdasarkan Karakteristik Kulit Dan Klasifikasi Sentimen Pada Platform Female Daily Menggunakan Metode LSTM

Tumanggor, Ave Regina S. U. D. R. (2026) Pengelompokan Produk Skincare Berdasarkan Karakteristik Kulit Dan Klasifikasi Sentimen Pada Platform Female Daily Menggunakan Metode LSTM. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003221190_Undergraduate_Thesis.pdf Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Perkembangan teknologi digital membuat penggunaan e-commerce meningkat, termasuk dalam pembelian produk kosmetik yang juga berkembang pesat di Indonesia. Female Daily merupakan salah satu platform ulasan kecantikan populer yang memuat banyak ulasan pengguna terhadap berbagai produk skincare. Namun, tidak semua ulasan sesuai antara penilaian bintang dan isi komentar, sehingga analisis sentimen diperlukan untuk memahami kecenderungan ulasan konsumen. Penelitian ini bertujuan untuk mengklasifikasikan sentimen ulasan produk skincare menggunakan metode Long Short-Term Memory (LSTM) yang kemudian digunakan sebagai dasar rekomendasi penggunaan produk berdasarkan tipe kulit pengguna. Data yang digunakan berupa teks ulasan berbahasa Indonesia dari Female Daily pada tiga merk yaitu S, G, dan W, yang dikategorikan menjadi sentimen positif dan negatif melalui pelabelan berbasis leksikon. Setelah melalui tahap preprocessing dan pembobotan kata menggunakan Word2Vec, model LSTM diterapkan untuk klasifikasi sentimen dan menghasilkan kinerja yang baik dengan nilai AUC sebesar 0,9696 pada data training dan 0,8131 pada data testing. Hasil klasifikasi sentimen kemudian dikaitkan dengan karakteristik tipe kulit pengguna melalui perhitungan proporsi tipe kulit pada tiap produk untuk menghasilkan rekomendasi produk yang cocok dan kurang cocok pada setiap tipe kulit. Hasil pengelompokan menunjukkan bahwa produk yang cocok lebih banyak dibandingkan yang kurang cocok pada setiap tipe kulit, dengan tipe kulit kombinasi memiliki jumlah produk cocok terbanyak. Temuan penelitian menunjukkan bahwa metode LSTM dapat mengidentifikasi sentimen dengan baik dan memberikan dasar rekomendasi pemakaian produk skincare yang sesuai dengan karakteristik kulit pengguna.
======================================================================================================================================
The development of digital technology has increased the use of e-commerce, including in the purchase of cosmetic products, which has also grown rapidly in Indonesia. Female Daily is one of the popular beauty review platforms that contains numerous user reviews of various skincare products. However, not all reviews are consistent between the star rating and the content of the comments, so sentiment analysis is needed to understand the tendency of consumer reviews. This study aims to classify the sentiment of skincare product reviews using the Long Short-Term Memory (LSTM) method, which is then used as a basis for product usage recommendations according to users' skin types. The data used consists of Indonesian-language review texts from Female Daily for three brands, namely S, G, and W, which are categorized into positive and negative sentiments through lexicon-based labeling. After undergoing the preprocessing stage and word weighting using Word2Vec, the LSTM model was applied for sentiment classification and produced good performance with an AUC value of 0.9696 on the training data and 0.8131 on the testing data. The sentiment classification results were then linked to users' skin type characteristics through the calculation of skin type proportions for each product to generate recommendations of suitable and less suitable products for each skin type. The grouping results show that suitable products are more numerous than less suitable ones for each skin type, with combination skin having the highest number of suitable products. The findings of this study indicate that the LSTM method can identify sentiment well and provide a basis for recommending the use of skincare products that match users' skin characteristics.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Analisis Sentimen, LSTM, Female Daily, Skincare, Tipe Kulit, Rekomendasi Produk ==================================================================================================================================================================================== Sentiment Analysis, LSTM, Female Daily, Skincare, Skin Type, Product Recommendation
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA278.5 Principal components analysis. Factor analysis. Correspondence analysis (Statistics) Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Ave Regina S. U. D. R Tumanggor
Date Deposited:	05 Aug 2026 03:24
Last Modified:	05 Aug 2026 03:24
URI:	http://repository.its.ac.id/id/eprint/143830

Actions (login required)

View Item