Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal TIN: TERAPAN INFORMATIKA NUSANTARA
Daniel Udjulawa
Universitas Multi Data Palembang, Palembang
Aerospace Engineering Arts Automotive Engineering Biochemistry, Genetics & Molecular Biology Chemical Engineering, Chemistry & Bioengineering Civil Engineering, Building, Construction & Architecture Computer Science & IT Decision Sciences, Operations Research & Management Dentistry Earth & Planetary Sciences Economics, Econometrics & Finance Energy Engineering Environmental Science Health Professions Immunology & microbiology Industrial & Manufacturing Engineering Law, Crime, Criminology & Criminal Justice Materials Science & Nanotechnology Mathematics Mechanical Engineering Medicine & Pharmacology Neuroscience Nursing Physics Social Sciences Veterinary Other

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Mitigasi Bias Feedback Loops dalam Rekomendasi Buku Menggunakan Pendekatan Causal Adjustment Yohanes Andika Dharma; Daniel Udjulawa
TIN: Terapan Informatika Nusantara Vol 6 No 12 (2026): May 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/tin.v6i12.9955

Abstract

The digital publishing industry has experienced exponential growth over the past decade, with platforms such as Goodreads and Amazon cataloging over 50 million book titles available online (Ricci et al., 2022). This abundance of choices paradoxically creates difficulty for users to discover books that genuinely match their preferences. The book domain was specifically chosen due to its unique characteristics: a highly asymmetric consumption distribution (bestsellers dominate 80% of sales despite representing only 5% of titles), extreme genre and language diversity, and readers' need for intellectual exploration beyond mere popularity. Collaborative filtering-based recommender systems address this challenge but are vulnerable to feedback loops that reinforce popularity bias, causing popular books to receive excessive exposure while long-tail items are neglected. This problem is exacerbated by Missing Not at Random (MNAR) data. This study proposes CAFL-SVD, a Matrix Factorization model based on SVD integrated with the CAFL algorithm through IPS and K-Means cluster regularization. Evaluated on Book-Crossing dataset with 585,579 ratings, CAFL-SVD reduces Gini coefficient by 37.7% (0.5783 to 0.3601), achieves peak NDCG@5 of 0.6207, maintains 100% coverage, and average Novelty Score of 14.0, demonstrating that causal approaches can simultaneously improve recommendation fairness and relevance without significant accuracy sacrifice.
Klasifikasi Penyakit Daun Kentang Berbasis CNN MobileNetV2 dengan Optimasi Randomize Search Jeason Lie; Abdul Rahman; Daniel Udjulawa
TIN: Terapan Informatika Nusantara Vol 6 No 9 (2026): February 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/tin.v6i9.9108

Abstract

Potatoes are a vital food commodity in Indonesia, but their productivity often declines significantly due to attacks by various leaf diseases that inhibit growth. This study aims to build an efficient and accurate automatic classification model for potato leaf diseases using Deep Learning technology. The approach used in this study is the MobileNetV2 Convolutional Neural Network (CNN) architecture based on transfer learning, which is known to have high computational efficiency. To obtain the most optimal model performance, this study applies an automatic hyperparameter tuning strategy using the Randomize Search method and performs robust model validation using the K-Fold Cross Validation technique with 5 folds. In addition, the balanced class weight technique is also applied to overcome the problem of data imbalance in the eight disease classes tested. The experimental results show that the best model configuration is achieved at the 15th iteration of the 5th fold using a combination of RMSprop optimizer parameters, 35 epochs, and a learning rate of 0.001. The final evaluation on independent test data produces an accuracy of 78.45%, a precision of 84.24%, and a recall of 72.94%. The very small difference between the validation accuracy of 79.30% and the test accuracy indicates good generalization ability without overfitting. Although the model achieved excellent results in Alternaria classification, challenges remained in identifying the Fungi class, which has high visual similarity to Phytopthora and Pest. This study concludes that the integration of MobileNetV2 with hyperparameter optimization is capable of effectively classifying potato leaf diseases.
Co-Authors Abdul Rahman Jeason Lie Yohanes Andika Dharma