Article Per Year (5 Year)
p-Index From 2021 - 2026
0.882
P-Index
This Author published in this journals
All Journal Sistemasi: Jurnal Sistem Informasi RABIT: Jurnal Teknologi dan Sistem Informasi Univrab Jurnal Sistem Komputer dan Informatika (JSON) Journal of Innovation Information Technology and Application (JINITA) Jurnal Algoritma
Giat Karyono
Universitas Amikom Purwokerto
Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Electrical & Electronics Engineering Engineering

Published : 5 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 5 Documents
Search

 1 
Visual Content Captioning and Audio Conversion using CNN-RNN with Attention Model Aldy Agil Hermanto; Giat Karyono; Imam Tahyudin; Boby Sandityas Prahasto
Journal of Innovation Information Technology and Application (JINITA) Vol 7 No 1 (2025): JINITA, June 2025
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v7i1.2788

Abstract

The primary objective of this research is to develop an image captioning and audio conversion system based on Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) with the integration of an Attention Mechanism, aimed at improving accessibility for visually impaired individuals. The research design follows a systematic approach involving data collection, preprocessing, model development, training, evaluation, and implementation. The methodology utilizes CNN for visual feature extraction, RNN for language modeling, and an Attention Mechanism to enhance contextual relevance in caption generation. Google Text-to-Speech (gTTS) is also integrated to convert generated captions into audio format. The main outcomes demonstrate that the model is capable of generating coherent and contextually relevant captions, as validated through qualitative assessment and quantitative measurement using the BLEU score. Experimental results show decreasing training and validation loss over 8 epochs without signs of overfitting, indicating stable model performance. The attention visualization confirms the model’s ability to focus on relevant image regions during caption generation. In conclusion, the proposed CNN-RNN architecture with Attention effectively generates descriptive captions and converts them into speech, showing strong potential for real-world accessibility applications.
Pengembangan Sistem Informasi Logistik Berbasis Web Dengan Fitur BAST Menggunakan Metode Waterfall Hamid Abdul Rozak; Giat Karyono
Jurnal Algoritma Vol 23 No 1 (2026): Jurnal Algoritma
Publisher : Institut Teknologi Garut

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33364/algoritma/v.23-1.3468

Abstract

The logistics information system at Universitas Amikom Purwokerto does not yet provide a digital Handover Report (BAST) feature. This study develops a BAST module integrated into a web-based logistics system using the Laravel framework and MySQL through the Waterfall method. The BAST feature is designed to be integrated with goods transaction data through a transactional checklist mechanism, allowing users to prepare, save, and print documents without manually re-entering data. Testing was conducted through Black Box Testing with 7 test scenarios and User Acceptance Testing (UAT) involving 2 users, covering aspects of functionality, ease of use, and document printing efficiency. All test scenarios were successfully executed according to requirements, and the system received good to very good ratings across all UAT aspects, indicating that the integration of the BAST feature can significantly improve the efficiency of logistics administration. Future research is recommended to use a more standardized usability measurement method, such as the System Usability Scale (SUS).
Systematic Literature Review Penerapan Physics-Informed Machine Learning untuk Analisis Seismisitas dan Bahaya Gempa Bumi Abdul Hakim Prima Yuniarto; Fandy Setyo Utomo; Giat Karyono
Jurnal Sistem Komputer dan Informatika (JSON) Vol. 7 No. 4 (2026): Juni 2026
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/json.v7i4.9830

Abstract

Pendekatan machine learning konvensional yang murni berbasis data dalam bidang seismologi sering kali menghasilkan model kotak hitam yang melanggar hukum fisika dasar serta memerlukan data pelabelan masif. Sebagai solusi, paradigma Physics-Informed Machine Learning (PIML) hadir mengintegrasikan pengetahuan domain fisis ke dalam arsitektur kecerdasan buatan. Penelitian ini bertujuan melakukan tinjauan literatur sistematis (Systematic Literature Review/SLR) untuk memetakan mekanisme integrasi, klaster parameter, dampak komputasi, serta kesenjangan riset dari penerapan PIML pada analisis kegempaan. Melalui evaluasi terstruktur terhadap 36 studi utama rentang 2021–2026, hasil analisis menunjukkan bahwa integrasi hukum fisika dominan dilakukan melalui modifikasi fungsi kerugian memanfaatkan residu persamaan diferensial parsial. Penerapan paradigma ini pada sejumlah studi terbukti mampu meningkatkan efisiensi komputasi serta mempercepat waktu simulasi numerik dibandingkan dengan metode konvensional, serta menyajikan pemodelan tanpa jaring yang stabil pada frekuensi tinggi. Namun, analisis kesenjangan mengungkap adanya ketergantungan pada skema terpandu deterministik. Riset ini merekomendasikan arah pengembangan masa depan pada integrasi metode unsupervised dan konstrain statistik seismologi guna menjamin kemasukakalan fisis model.
DETEKSI ANOMALI PADA AUDIT BARANG MILIK DAERAH MENGGUNAKAN EXTREME GRADIENT BOOSTING TEROPTIMASI BAYESIAN Sulman Edi S; Giat Karyono; Imam Tahyudin
Rabit : Jurnal Teknologi dan Sistem Informasi Univrab Vol 11 No 2 (2026): Juli
Publisher : LPPM Universitas Abdurrab

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.36341/rabit.v11i2.8285

Abstract

The audit and reconciliation of Local Government-Owned Assets (BMD) constitute a critical governance process that remains heavily reliant on manual verification, rendering it susceptible to inefficiency and human error, particularly within datasets exhibiting extreme heterogeneity in acquisition values. This study proposes an intelligent analytical framework for asset depreciation anomaly detection using the Extreme Gradient Boosting (XGBoost) algorithm, optimized via a Bayesian Optimization approach through the Optuna framework. From an initial raw population of 98,526 records, the data underwent preprocessing to yield 42,241 clean records with unique profiles. To address the disparity in price ranges, the dataset was divided into three strata using the Equal Frequency quantile method, with the prediction target transformed into a depreciation ratio. The evaluation demonstrated highly precise performance, consistently achieving a coefficient of determination (R²) above 0.99. Bayesian optimization reduced the Weighted Average Percentage Error (WAPE) to a range of 0.54% to 1.32%. Using a 10% deviation threshold, the system automatically extracted 101 anomalous records (1.20%) from 8,450 test samples as red flags. The results confirm that this framework is highly viable as a decision-support instrument for public asset audits, in compliance with regional regulations.
WhatsApp Hybrid Chatbot Architecture Rasa-DeepSeek: Design and Performance Evaluation Iqbaluddin Syam Had; Fandy Setyo Utomo; Giat Karyono; Dwi Putriana Nuramanah Kinding
SISTEMASI Vol 15, No 2 (2026): Sistemasi: Jurnal Sistem Informasi
Publisher : Universitas Islam Indragiri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32520/stmsi.v15i2.5791

Abstract

This study designed and evaluated a hybrid chatbot for a domain-specific application by addressing two main issues: limited NLU coverage and the variability of latency and cost when all queries are routed directly to an LLM. The proposed solution integrates a deterministic Rasa-based pipeline with a DeepSeek fallback mechanism. In this architecture, Rasa handles NLU processing, rules, stories, and context storage for mk and jk, while the LLM is only invoked when the NLU confidence score falls below a defined threshold. The methodology includes end-to-end implementation through a Node.js bridge connected to Rasa, functional testing to validate the intent–entity–action flow, and performance testing using load (stress) testing across two access paths: the Rasa REST endpoint and the Node-to-Rasa bridge. Meanwhile, the LLM pipeline was profiled separately through instrumented action calls. The results indicate that domain-specific conversations were successfully answered using curated knowledge, and both deterministic access paths met the service level objective (SLO), achieving a median latency of approximately 32 milliseconds with no observed errors. This study contributes by demonstrating that a hybrid chatbot architecture separating deterministic and generative pipelines can maintain SLO compliance in domain-specific settings. In addition, it highlights limitations of LLMs in understanding domain ontologies, reinforcing the need for semantic guardrails.
Co-Authors Abdul Hakim Prima Yuniarto Aldy Agil Hermanto Boby Sandityas Prahasto Dwi Putriana Nuramanah Kinding Fandy Setyo Utomo Hamid Abdul Rozak Imam Tahyudin Iqbaluddin Syam Had Sulman Edi S