Klasterisasi Kategori Judul Buku Pada Perpustakaan Dengan Menggunakan Metode HDBSCAN

Laksono, Ivan Luthfi
Pribadi, Muhammad Rizky

Algoritme Jurnal Mahasiswa Teknik Informatika
Vol 6 No 2 (2026): April 2026 || Algoritme Jurnal Mahasiswa Teknik Informatika

To assist in library collection management, this study aims to create a system that can automatically classify book titles. Previous studies have mostly used K-Means and DBSCAN because they have limitations in determining the number of clusters and are less responsive to varying densities of text data. Furthermore, HDBSCAN is still limited to clustering Indonesian-language book titles. The dataset consists of 1044 book titles that were processed through text preprocessing, TF-IDF weighting, and dimension reduction using Singular Value Decomposition (SVD). When HDBSCAN was used for clustering and compared with DBSCAN, the results showed that the combination of SVD and HDBSCAN had better cluster quality with a Silhouette value of 0.158 and a lower noise level. This study scientifically demonstrates that improving the stability of cluster structures in large book title data can be achieved through the integration of dimension reduction and density-based clustering.