Sistem Rekomendasi Ayat Al-Qur'an Berbasis K-Means Clustering dan Cosine-Similarity dengan Reduksi Dimensi PCA

Akbar, Amarul (2026) Sistem Rekomendasi Ayat Al-Qur'an Berbasis K-Means Clustering dan Cosine-Similarity dengan Reduksi Dimensi PCA. Masters thesis, Univeritas Pembangunan Nasional (Veteran) Jawa Timur.

	Text (COVER) 23066020015-cover.pdf Download (7MB)
	Text (BAB 1) 23066020015-bab1.pdf Download (145kB)
	Text (BAB 2) 23066020015-bab2.pdf Restricted to Repository staff only until 5 February 2028. Download (746kB)
	Text (BAB 3) 23066020015-bab3.pdf Restricted to Repository staff only until 5 February 2028. Download (1MB)
	Text (BAB 4) 23066020015-bab4.pdf Restricted to Repository staff only until 5 February 2028. Download (9MB)
	Text (BAB 5) 23066020015-bab5.pdf Download (100kB)
	Text (DAFTAR PUSTAKA) 23066020015-daftarpustaka.pdf Download (112kB)
	Text (LAMPIRAN) 23066020015-lampiran.pdf Restricted to Repository staff only until 5 February 2028. Download (481kB)

Abstract

The development of information technology has facilitated access to the Qur’an through various digital platforms; however, the large number of verses and the diversity of themes often make it difficult for users to find verses that are relevant to specific needs. This study aims to develop a Qur’anic verse recommendation system based on Indonesian translated text using a quantitative approach. The system is built using TF-IDF as text representation, K-Means Clustering to construct a global lexical–thematic structure of verses, and Cosine Similarity as a ranking mechanism based on the similarity of user keywords. To address the high dimensionality of the data resulting from TF-IDF vectorization, Principal Component Analysis (PCA) is applied as a dimensionality reduction technique prior to the clustering process. PCA serves as a supporting method to simplify the feature space and relatively improve cluster quality. The evaluation is conducted quantitatively using Silhouette Score (SS) and Davies-Bouldin Index (DBI) to measure cluster quality, as well as precision to assess recommendation performance. The experimental results indicate that the application of PCA improves cluster quality compared to the model without dimensionality reduction, although the Silhouette Score remains in the moderate category. The recommendation system achieves a precision value of 76% in the Top-5 recommendation scenario. Although the system is still based on lexical similarity and has not yet captured deep semantic meaning, the integration of K-Means Clustering, Cosine Similarity, and PCA is proven to produce a relevant and efficient Qur’anic verse recommendation system to support verse retrieval for learning and contemplative (tadabbur) purposes.

Item Type:

Thesis (Masters)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Rahmat, Basuki	NIP196907232021211002	basukirahmat.if@upnjatim.ac.id
Thesis advisor	Sari, Anggraini Puspita	NPT22219860816400	anggraini.puspita.if@upnjatim.ac.id

Uncontrolled Keywords:

Sistem Rekomendasi, Al-Qur’an, K-Means Clustering, Cosine Similarity, Principle Component Analysis, Precision.

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Magister Information Technology

Depositing User:

AMARUL AMAR AKBAR

Date Deposited:

09 Feb 2026 07:10

Last Modified:

09 Feb 2026 07:10