Sistem Rekomendasi Buku Bacaan untuk Anak Menggunakan Collaborative Filtering dan Topic Modelling

Priady, Farrel Edgarrafi (2024) Sistem Rekomendasi Buku Bacaan untuk Anak Menggunakan Collaborative Filtering dan Topic Modelling. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003201160-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (3MB) | Request a copy

Abstract

Membaca buku merupakan salah satu kegiatan penting dalam perkembangan anak, terutama pada usia emas (0 – 6 tahun). Namun, di era digital, anak-anak dihadapkan dengan berbagai rangsangan yang kuat, sehingga penting untuk memberikan bacaan yang sesuai dengan minat dan usia mereka. Pengembangan sistem rekomendasi buku bacaan untuk anak menjadi suatu kebutuhan mendesak guna meningkatkan minat dan kualitas literasi anak di Indonesia. Penelitian ini bertujuan untuk membangun sistem rekomendasi buku bacaan untuk anak menggunakan dua pendekatan, yaitu collaborative filtering dan topic modelling. Data yang digunakan adalah data judul, deskripsi, dan rating buku yang diambil dari website Goodreads yang disediakan oleh University of California San Diego (UCSD). Berdasarkan hasil penelitian yang telah dilakukan menggunakan Grid Search Cross Validation dengan 5 fold, didapatkan bahwa model terbaik adalah sistem rekomendasi menggunakan faktorisasi matriks SVD dengan nilai evaluasi dari model tersebut adalah RMSE sebesar 0,7941, accuracy sebesar 79,89%, dan F1-Score sebesar 88,28%. Model tersebut lebih baik daripada metode pembanding yaitu LDA First dengan nilai evaluasi dari model tersebut adalah RMSE sebesar 0,9011, accuracy sebesar 78,31%, dan F1-Score sebesar 87,81%. Penelitian selanjutnya disarankan untuk melakukan hibridisasi atau penggabungan dari metode SVD dan LDA tersebut secara bersamaan, juga menambah data pengguna seperti umur, jenis kelamin, atau lokasi dari pengguna untuk menangani cold start.
==========================================================================================================================
Reading books is one of the most important activities in children's development, especially at the golden age (0 – 6 years old). However, in the digital age, children are faced with a variety of powerful stimuli, so it is important to provide reading that matches their interests and age. The development of a system of reading book recommendations for children has become an urgent need to increase the interest and quality of children's literacy in Indonesia. The research aims to build a system of reading book recommendations for children using two approaches, which is collaborative filtering and topic modelling. The data used is book rating data taken from the Goodreads website provided by the University of California San Diego (UCSD). Based on the results of research conducted using Grid Search Cross Validation with 5 folds, it was found that the best model was a recommendation system using SVD matrix factorization, with the evaluation value of the model being RMSE of 0,7941, accuracy of 79,89%, and F1-Score of 88,28%. The model was better than the comparison method, the LDA First with the evaluation value of the model was an RMSE of 0,9011, an accuracy of 78,31%, and F1-score of 87,81%. Further research is recommended to perform hybridization or a combination of the SVD and LDA methods, as well as adding the user's age, gender, or location to handle cold start.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Buku Bacaan Anak, Faktorisasi Matriks SVD, Sistem Rekomendasi, Topic Modelling LDA Children’s Book, LDA Topic Modelling, Recommender Systems, SVD Matrix Factorization.
Subjects:	H Social Sciences > HA Statistics Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA76.9.I58 Recommender systems (Information filtering)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Farrel Edgarrafi Priady
Date Deposited:	08 Aug 2024 12:20
Last Modified:	08 Aug 2024 12:20
URI:	http://repository.its.ac.id/id/eprint/114962

Actions (login required)

View Item