cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
Muhammad Nur Faiz
Contact Email
faiz@pnc.ac.id
Phone
+6282324039994
Journal Mail Official
jinita.ejournal@pnc.ac.id
Editorial Address
Department of Informatics Engineering Politeknik Negeri Cilacap Jln. Dr.Soetomo No.01 Sidakaya, Cilacap, Indonesia
Location
Kab. cilacap,
Jawa tengah
INDONESIA
Journal of Innovation Information Technology and Application (JINITA)
Published by Politeknik Negeri Cilacap
ISSN : 27160858     EISSN : 27159248     DOI : https://doi.org/10.35970/jinita.v2i01.119
Core Subject : Science, Engineering,
Computer Science & IT Decision Sciences, Operations Research & Management Engineering
Software Engineering, Mobile Technology and Applications, Robotics, Database System, Information Engineering, Interactive Multimedia, Computer Networking, Information System, Computer Architecture, Embedded System, Computer Security, Digital Forensic Human-Computer Interaction, Virtual/Augmented Reality, Intelligent System, IT Governance, Computer Vision, Distributed Computing System, Mobile Processing, Next Network Generation, Natural Language Processing, Business Process, Cognitive Systems, Networking Technology, and Pattern Recognition
Arjuna Subject : Ilmu Komputer (semua kategori) - Ilmu Komputer (Lain-Lain)
Articles 191 Documents
 16   17   18   19   20 
Demystifying Political Hate Speech Detection: An Explainable Artificial Intelligence Audit of Transformer and Linear Models Aulia Miftah Razak; Shofwatul Uyun; Abdul Rozak
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3205

Abstract

The rise of political hate speech on social media calls for detection systems that are not only accurate but also transparent to avoid moderation bias. Although Transformer models achieve high performance, their “black box” nature creates the risk of a “false sense of security” in content moderation, where high accuracy can mask systemic bias. This study aims to transparently audit the decision-making mechanisms of a lexical-feature-based Support Vector Machine (SVM) model and a contextual-representation-based IndoBERT model using an Explainable AI (XAI) approach via the LIME method. Experimental results show that IndoBERT significantly outperforms SVM with a Macro-F1 score of 90.8% versus 84.0%. However, the XAI audit revealed the presence of data-driven bias in both models toward specific political entities such as “Jokowi,” “Prabowo,” “cebong,” and “kampret,” which often trigger negative labels automatically without a comprehensive contextual review. These findings underscore that transparency audits through XAI serve as a crucial bridge for building a content moderation system that is fair, accountable, and capable of protecting freedom of expression within the digital democratic ecosystem.
A Reproducible Explainable NLP Workflow for Workplace Sexism Detection: Classification Performance, Rationale Faithfulness, and Sanity Checks Annisa Romadloni; Linda Perdana Wanti; Laura Sari; Muhammad Nur Faiz; Qisthi Alhazmi Hidayaturrohman
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3222

Abstract

Workplace sexism often appears as indirect, deniable language (e.g., patronizing compliments, competence-doubting questions), making automated detection and organizational response difficult. This study evaluates a transparent, explanation-ready NLP pipeline on the Sexist Workplace Statements (SWS) dataset (1,137 items) with its binary labels: certain sexism vs. ambiguous/neutral. Using the provided fixed stratified split (1,023 train; 114 test), we train a TF–IDF (word 1–2, character 3–5 n-grams) logistic regression baseline and report performance stability across five random seeds. To audit model evidence, sparse token rationales are extracted from linear feature contributions and quantify faithfulness with ERASER-style comprehensiveness (logit drop when rationales are removed) and sufficiency (logit change when only rationales are kept), benchmarked against random-token rationales. The baseline achieves 0.768 ± 0.006 accuracy and 0.759 ± 0.007 macro-F1, with errors concentrated in the ambiguous/neutral class. Faithfulness tests show that model-selected rationales substantially affect the sexism logit (comprehensiveness 1.335 ± 0.001), while remaining insufficient in isolation (|sufficiency| 1.075 ± 0.006). Sanity checks reveal modest sensitivity to gender-term swaps and reduced rationale overlap underweight randomization. Overall, results motivate cautious deployment: explanation-driven auditing can surface shortcut risks and clarify where binary labels blur neutral language and deniable sexism, pointing to future work on finer-grained annotation and human rationale collection.
Edge Detection and Post-Edge Filtering for Crescent Detection Using Circular Hough Transform Ady Suprayitno; Murinto; Kartika Firdausy
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3229

Abstract

Crescent detection based on digital images faces major challenges due to low contrast, atmospheric noise, and the very thin curved structure of the crescent moon. This study aims to evaluate the influence of edge detection and post-edge filtering methods on the performance of crescent moon detection based on the Circular Hough Transform (CHT). Three edge detection methods, namely Sobel, Canny, and Phase Stretch Transform (PST), were combined with three filtering techniques, namely Gaussian, Median, and Bilateral filtering. All images were processed using a fixed pre-processing pipeline consisting of image resizing, conversion to grayscale, and contrast enhancement using CLAHE. Performance evaluation was conducted using Accuracy, Precision, Recall, and F1-Score as the main metrics, and further verified through visual analysis of CHT detection results. The results show that Canny achieves the best performance in unfiltered conditions, while filtering significantly improves detection reliability depending on the characteristics of the edge detector. The combination of Canny and Bilateral filtering produces the highest accuracy and results in the most geometrically consistent CHT detection. On the other hand, Median filtering was found to be unsuitable for Canny due to its tendency to remove thin edge structures that are critical for crescent detection. These findings highlight the crucial role of edge map quality in CHT-based crescent detection and provide methodological guidance for selecting reliable combinations of edge detection and filtering in automated crescent observation systems.
Rice Planting Time Prediction Using SARIMA-MFEP Integration in Kubu Raya Sinta Rama Dani; Syarifah Putri Agustini Alkadri; Sucipto
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3233

Abstract

Extreme climate change has increased uncertainty in rice planting schedules, threatening food security in Kubu Raya Regency, West Kalimantan, and causing significant economic losses due to inaccurate seasonal predictions. This study integrates the Seasonal Autoregressive Integrated Moving Average (SARIMA) method with the Multi-Factor Evaluation Process (MFEP) to generate rice planting time recommendations based on scientific climate forecasting and multi-criteria agroclimatic evaluation. SARIMA is employed to forecast monthly rainfall, temperature, and humidity, while MFEP evaluates the feasibility of twelve alternative planting months using weighted criteria determined by local agricultural experts. The objective of this research is to develop an objective, accurate, and validated planting time prediction system to support farmers’ decision-making. The results show that the SARIMA model achieves very high accuracy, with Mean Absolute Percentage Error (MAPE) values below 2% for both temperature and humidity, and successfully captures 68% of seasonal rainfall variability. October is identified as the optimal planting month with the highest feasibility score, consistent with historical peak harvest patterns in January and February and aligned with regional literature. This integrated approach provides an end-to-end solution from forecasting to empirically validated, actionable recommendations, offering strong potential to reduce crop failure risk and enhance rice production efficiency under climate uncertainty.
Development of Prototype IoT-Based Control and Monitoring System for Pakcoy Better Swengky; Lesta; Peprizal; Rian Ardiansyah; Yosar Leman; Aldian Kurniawan
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3247

Abstract

This research aims to design and develop an Internet of Things (IoT)-based automatic control and monitoring system for hydroponic plants. This system measures, displays, and regulates nutrient parameters, pH, and water volume in real-time to improve water use efficiency, nutrient stability, and plant productivity. The method used is Research and Development (R&D) with a prototyping model. This system integrates an ESP32 microcontroller with a DFRobot pH sensor, a TDS (Total Dissolved Solids) sensor, and an ultrasonic sensor to monitor water levels. Data from the sensors is processed and sent to a cloud database to be monitored and controlled through a website interface. The test results show that the system has high accuracy, with a MAPE (Mean Absolute Percentage Error) value for the pH sensor of 6.06% (93.94% accuracy), a TDS sensor of 4.81% (95.19% accuracy), and an ultrasonic sensor of 7.98% (92.02% accuracy). All MAPE values ​​are within the <10% range, indicating good system performance for automated monitoring applications. This system implementation provides an innovative solution to the limitations of conventional, manual hydroponic systems. It can serve as an initial model for the future development of IoT-based precision agriculture.
A Hybrid Ensemble Learning Framework for Forecasting Renewable Energy Stocks: Evidence from the Indonesia Stock Exchange Hana A. R.; Hapsari Peni Agustin Tjahyaningtijas
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3274

Abstract

Forecasting stock prices in the renewable energy sector remains a challenging task due to high market volatility, regulatory dynamics, and rapidly changing investor sentiment. Accurate prediction models are therefore essential to support informed investment decisions and reduce financial uncertainty. This study proposes a hybrid ensemble forecasting framework that integrates Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Support Vector Regression (SVR) to improve the prediction accuracy of renewable energy stock prices. The model leverages the complementary strengths of LSTM in capturing long-term temporal dependencies and GRU in efficiently modeling short-term market dynamics, while SVR acts as a meta-learner within a stacking ensemble architecture to enhance predictive generalization. To further strengthen the predictive capability, a set of technical indicators is incorporated to enrich the feature representation, and Bayesian optimization is employed to adaptively tune key hyperparameters. The model is evaluated using renewable energy stocks listed on the Indonesia Stock Exchange with hourly trading data, and its performance is assessed using multiple regression metrics, including Mean Squared Error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and the coefficient of determination (R²). Experimental results demonstrate that the proposed hybrid model consistently outperforms several benchmark approaches, achieving superior forecasting accuracy and more stable predictions across multiple stocks. These findings indicate that the integration of complementary recurrent architectures with ensemble learning and adaptive optimization provides a robust framework for modeling complex financial time-series data, offering improved predictive reliability for renewable energy stock forecasting.
User Sentiment toward Halodoc Psychological Consultation Services Using RNN and LSTM Analysis Riyadi Purwanto; Randi Afif; Fajar Mahardika; Dwi Novia Prasetyanti; Cahya Vikasari; Rostika Listyaningrum
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3288

Abstract

In the digital era, online psychological consultation services such as HaloDoc are increasingly used to provide accessible mental health support; however, the rise in users does not always reflect service satisfaction, making sentiment analysis of user reviews essential for understanding public perception. This study aims to classify positive and negative sentiments from Indonesian-language user reviews of HaloDoc’s psychological consultation service and to compare the performance of Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) algorithms. Using a quantitative approach with text mining and sentiment analysis, data were collected from Google Play Store reviews and processed through case folding, filtering, tokenizing, and stemming, then manually labeled and split into 80% training and 20% testing data. Models were developed using Keras with RNN and LSTM architectures and evaluated using confusion matrix metrics, including accuracy, precision, recall, and F1-score. The results show that RNN achieved 95% accuracy, while LSTM reached 91%; although RNN performed better in accuracy, LSTM demonstrated more stable performance and superior capability in capturing complex contextual information, particularly in longer reviews with varied emotional expressions. Overall, the findings indicate that LSTM is more effective for Indonesian-language sentiment classification in the digital psychology domain and can serve as a foundation for developing automated systems to assess user satisfaction with online psychological services in Indonesia
Robust DeBERTa-v3 Framework for Aspect-Based Sentiment Analysis with Extreme Class Imbalance Muhammad Rikzam Kamal; Sabrina Ahmad; Rohmad Abidin; Imam Prayogo Pujiono
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3296

Abstract

Aspect-Based Sentiment Analysis (ABSA) often operates under extreme class imbalance, causing Transformer-based models to favour majority classes while failing to detect minority sentiments disproportionately. This study proposes RoABSA, a robust DeBERTa-v3–based framework that integrates Hybrid Semantic Augmentation, combining Easy Data Augmentation and Back-Translation, with cost-sensitive optimization via Weighted Random Sampling and Focal Loss to enhance diversity representation and recalibrate gradient contributions. Evaluated on four SemEval benchmarks (Lap14, Res14, Res15, Res16), RoABSA achieves consistent state-of-the-art performance with Macro-F1 scores of 84.71, 91.65, 88.40, and 84.24, respectively, outperforming strong graph-based and Transformer-based baselines by margins of up to 10.35 points. Ablation results confirm that robustness emerges from the synergy between augmentation and cost-sensitive learning rather than any single component. At the same time, per-class analysis demonstrates substantial gains in detecting minority sentiments, including cases where baselines failed. These findings highlight that addressing long-tailed distributions in ABSA requires coordinated interventions at both data and optimization levels, and that RoABSA provides a practical, generalizable strategy for improving stability and fairness in sentiment classification.
A Multimodal Deep Learning Framework for Herbal Plant Classification with Large Language Model Integration Shalsabila Dwi Aprilia; Nunu Nurdiana; Harun Sujadi
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3297

Abstract

Accurate identification of herbal plants remains challenging due to high species diversity and the need for specialized botanical knowledge. This study proposes a multimodal artificial intelligence framework that integrates deep learning-based image classification with a large language model to provide both plant identification and descriptive information. The objective is to develop a more informative and practical system that extends beyond conventional classification approaches. The method employs a ResNet50V2 architecture with transfer learning to classify herbal plant images, combined with a large language model to automatically generate descriptions of plant benefits. The model was trained on 19,213 images across 131 plant classes. Experimental results demonstrate an accuracy of 89.43%, with a macro F1-score of 0.85 and a weighted F1-score of 0.89, indicating strong and consistent performance across classes. The integration of visual recognition and language generation enables the system to deliver richer, user-oriented outputs compared to existing methods that only provide classification labels. In addition, the system achieves real-time inference performance, making it suitable for web-based applications. This study contributes a scalable and comprehensive multimodal framework for herbal plant identification, highlighting the effectiveness of combining computer vision and large language models to enhance both accuracy and usability.
Random Forest-Based Multiclass Classification of Pestalotiopsis sp. Disease Severity in Rubber Plants Using UAV Multispectral Data. Solikin; Yeni Herdiyeni; Annisa; Lilik Budi Prasetyo; Tri Rapani Febbiyanti; Imas Sukaesih Sitanggang; Sri Nurdiati
Journal of Innovation Information Technology and Application (JINITA) Vol 8 No 1 (2026): JINITA, June 2026
Publisher : Politeknik Negeri Cilacap

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35970/jinita.v8i1.3310

Abstract

Pestalotiopsis sp.-induced leaf-fall disease threatens rubber (Hevea brasiliensis) productivity by reducing canopy density, degrading chlorophyll, and impairing photosynthetic capacity. This study developed an exploratory multiclass framework to classify disease severity in rubber plantations using UAV multispectral data and a Random Forest (RF) algorithm. The dataset comprised 144 ground-truth canopy-level samples labeled into four ordinal severity classes, L1 (mild), L2 (moderate), L3 (severe), and L4 (very severe), and 1760 additional UAV-derived observations for spatial inference. Spectral and structural predictors included the Leaf Chlorophyll Index (LCI), Normalized Difference Red Edge (NDRE), density_LCI, and density_NDRE, collected across three rubber clones, BPM 24, GT 1, and RRIC 100. Model development used an 80:20 stratified train–test split, stratified and nested cross-validation, class weighting, and SMOTE applied only within training folds to reduce imbalance and leakage. Performance was evaluated using accuracy, class-wise precision–recall–F1, and the multiclass Matthews Correlation Coefficient (MCC). Hyperparameter tuning (n_estimators = 150, max_depth = 7, min_samples_split = 9) increased MCC from 0.143 to 0.547 on independent test data and 0.586 under the all-labeled-data scenario, indicating moderate agreement. LCI emerged as the dominant predictor, supporting the physiological relevance of chlorophyll-sensitive red-edge information. However, low recall for L2 and unstable L4 performance reflected class imbalance, limited labeled samples, and overlapping spectral responses. The framework provides a more informative severity stratification than binary detection, but further ground-truth expansion, ordinal or cost-sensitive modeling, comparative benchmarking, and multi-season validation are required before operational deployment.

Page 19 of 20 | Total Record : 191

 16   17   18   19   20 

Filter by Year

2019 2026


Reset
Filter By Issues
All Issue Vol 8 No 1 (2026): JINITA, June 2026 Vol 7 No 2 (2025): JINITA, December 2025 Vol 7 No 1 (2025): JINITA, June 2025 Vol 6 No 2 (2024): JINITA, December 2024 Vol 6 No 1 (2024): JINITA, June 2024 Vol 5 No 2 (2023): JINITA, December 2023 Vol 5 No 1 (2023): JINITA, June 2023 Vol 4 No 2 (2022): JINITA, December 2022 Vol 4, No 1 (2022): JINITA, June 2022 Vol 4 No 1 (2022): JINITA, June 2022 Vol 3 No 1 (2021): JINITA, June 2021 Vol 3 (2021): JINITA, December 2021 Vol 2 No 2 (2020): JINITA, December 2020 Vol 2 No 1 (2020): JINITA, June 2020 Vol 1 No 1 (2019): JINITA, December 2019 More Issue