cover
Contact Name
muhammad siddik hasibuan
Contact Email
mhdsiddikhasibuan@gmail.com
Phone
-
Journal Mail Official
cosieaira@gmail.com
Editorial Address
Jl Pukat Banting IV NO 41 Medan Kecamatan Medan Tembung Kode Pos 20224
Location
Kota medan,
Sumatera utara
INDONESIA
Journal of Computer Science and Informatics Engineering
ISSN : -     EISSN : 28278356     DOI : -
Core Subject : Science,
Artificial Intelligence Machine Learning Natural Language Processing Computer Vision Text Speech Text Mining Data mining Cryptography Data visualization Expert System Deep Learning Fuzzy Logic IoT and smart environments Neural Networks Pattern Recognition Image Processing Optimization Digital Signal Processing Networking Technology Web intelligence
Articles 147 Documents
Comparative Analysis of MambaOut Tiny and Vision Transformer for Corn Leaf Disease Classification Adi Slamet Priyadi; Albert Yakobus Chandra
Journal of Computer Science and Informatics Engineering Vol 5 No 3 (2026): July
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55537/cosie.v5i3.1834

Abstract

Maize production remains vulnerable to foliar diseases such as Gray Leaf Spot, Common Rust, and Northern Leaf Blight, which can impair growth and reduce crop yields. Manual disease identification is time-consuming, labor-intensive, and reliant on the observer's expertise, necessitating automated detection systems based on deep learning. While Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) are widely used for plant disease classification, a direct comparison between ViT-B/16 and MambaOut Tiny under identical training conditions remains rare. This study compares the two models using the PlantVillage dataset, which comprises 4,188 maize leaf images across four classes. Both models were fine-tuned using ImageNet-1K pre-trained weights, the AdamW optimizer, a learning rate of 1e-4, a batch size of 16, and 10 epochs. The results demonstrate that MambaOut Tiny achieved an accuracy of 97% and a macro-average F1-score of 0.96, outperforming ViT-B/16, which achieved 95% accuracy and an F1-score of 0.93. Additionally, MambaOut Tiny features a lower parameter count (30 million versus 86 million) and faster inference time (0.01262 seconds per image). These findings indicate that MambaOut Tiny is more efficient for precision agriculture systems with limited computational resources
Comparative Analysis of TF-IDF, TF-IDF+WordNet, and Sentence-BERT for News Document Retrieval Using Cosine Similarity Galib Haftha Zuhayir; Wiwik Suharso; Nanda Kurnia Wardati
Journal of Computer Science and Informatics Engineering Vol 5 No 3 (2026): July
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55537/cosie.v5i3.1835

Abstract

The rapid growth of digital news volume has produced information overload, while exact keyword-matching retrieval remains vulnerable to synonymy and polysemy, causing relevant documents to be missed. Prior studies generally compare only two document representation methods on small-scale datasets, leaving a gap in controlled evaluations that jointly compare statistical, lexical-hybrid, and neural approaches on a large-scale news domain. This study compares three document representation methods, namely Term Frequency-Inverse Document Frequency (TF-IDF), TF-IDF with WordNet-based query expansion, and Sentence-BERT (all-MiniLM-L6-v2), for news document retrieval using Cosine Similarity on the BBC News Dataset (14,305 documents with a hierarchical Ground Truth of 5 Topics and 51 Subtopics). Ten queries were evaluated using Precision@K, Recall@K, F1-Score@K (K=5, 10, 20), and execution time. The results show that Sentence-BERT consistently outperforms the other methods with a Precision@5 of 0.84, compared to TF-IDF (0.56) and TF-IDF+WordNet (0.52), while TF-IDF remains the fastest at online query time (23.86 ms per query). WordNet expansion actually reduces precision and increases execution time without a proportional accuracy gain. These findings confirm that transformer-based semantic representations are superior for news domains with high lexical variation, while TF-IDF remains relevant for computationally constrained real-time systems
Comparison of SVM and Random Forest with RFE for Diabetes Prediction Anggi Vandryan; Putry Wahyu Setyaningsih
Journal of Computer Science and Informatics Engineering Vol 5 No 3 (2026): July
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55537/cosie.v5i3.1838

Abstract

Individuals with a high Body Mass Index (BMI) are among the most vulnerable groups to developing Type 2 Diabetes Mellitus due to insulin resistance caused by visceral fat accumulation. However, most existing machine learning models have been developed using general population data without considering the specific characteristics of high-risk individuals. This study aims to analyze and compare the performance of Support Vector Machine (SVM) and Random Forest (RF) algorithms in predicting diabetes risk among individuals with a BMI ≥ 25, while also evaluating the impact of Recursive Feature Elimination (RFE) on improving model performance. The Pima Indians Diabetes Dataset from the UCI Machine Learning Repository was used as the data source. After filtering records based on BMI, a total of 662 instances were included in the analysis. The preprocessing stage consisted of median imputation for invalid values, feature normalization using StandardScaler, and feature selection using RFE to select four features for each model. The dataset was divided into training and testing sets using a 70:30 ratio. Model performance was evaluated using accuracy, precision, recall, F1-score, and confusion matrix metrics. The results indicate that the RF-RFE model achieved the best performance, with an accuracy of 78%, a recall of 68% for the diabetes class, and an F1-score of 72%, representing a significant improvement over the RF model without RFE (74% accuracy and 60% recall). The combination of Random Forest and Recursive Feature Elimination proved to be the most effective approach for reducing false negatives, which is particularly important in the context of early clinical detection of diabetes
Integrating Word Embeddings and IMDb Web Scraping for Keyword-Based Movie Recommendation Andani Chacha Cahya Dewi; Deni Arifianto; Nanda Kurnia Wardati
Journal of Computer Science and Informatics Engineering Vol 5 No 3 (2026): July
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55537/cosie.v5i3.1839

Abstract

The rapid growth of the film industry and streaming platforms has led to information overload and filter bubbles that make it difficult for users to find content matching their narrative preferences. Prior content-based filtering approaches relying on word-frequency methods (TFIDF) suffer from a semantic gap and commonly depend on a single public dataset and a reference title (seed movie) as input. This study combines a public dataset with IMDb web-scraping results (a maximum population of 5,000 titles) and applies a Skip-gram Word2Vec model to represent movie synopses as 200-dimensional semantic vectors, paired with Cosine similarity to measure the closeness between a user's free-text keyword and movie synopses without requiring a seed movie. Data were split using an 80:20 Holdout method, and algorithm performance was evaluated on a Top-3 Recommendation window using Precision@K, Recall@K, and Mean Reciprocal Rank (MRR), with ground truth validated by two experts through Inter-Annotator Agreement. Testing on 25 queries produced a Precision@3 of 0.5333, Recall@3 of 0.7800, and MRR of 0.7300. These results indicate that integrating word embeddings with web scraping yields semantically relevant movie recommendations from free keyword input, though comparisons with baseline methods are needed for more definitive performance claims
Web-Based Information System Design for BUMDes Sakinah in Sialang Panjang Village Hasan Arya Sandy; Ilyas Ilyas; Abdul Muni
Journal of Computer Science and Informatics Engineering Vol 5 No 3 (2026): July
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55537/cosie.v5i3.1843

Abstract

Village-Owned Enterprises (BUMDes) play an important role in improving the rural economy. However, administrative management, business data processing, document archiving, and information dissemination at BUMDes Sakinah in Sialang Panjang Village were still carried out manually, resulting in inefficiency, a high risk of data loss, and delays in providing accurate information. This condition indicates a gap in the utilization of information technology to support integrated BUMDes management. This study aimed to design a web-based information system for BUMDes Sakinah as a solution to these problems. A qualitative research method was employed, with data collected through observation, interviews, and documentation. The system was developed using the System Development Life Cycle (SDLC), consisting of planning, analysis, design, implementation, testing, and maintenance stages. The results show that the proposed system successfully integrates business data management, digital document archiving, and information services into a single web-based platform, thereby improving effectiveness, efficiency, data security, and accessibility of information. Practically, the system supports better administrative management and service quality, while academically it contributes to the implementation of web-based information systems for Village-Owned Enterprises. In conclusion, the developed system has proven to enhance the effectiveness, transparency, and accountability of BUMDes management
Decision Support System for Medical Equipment Supplier Selection Recommendations Using MABAC and SAW Methods Annisa Zafhitri Nasution; Muhammad Dedi Irawan
Journal of Computer Science and Informatics Engineering Vol 5 No 3 (2026): July
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55537/cosie.v5i3.1873

Abstract

This study aims to develop a decision support system for selecting medical device suppliers at PT. Azkha Medical Indonesia using a combination of the Simple Additive Weighting (SAW) and Multi-Attributive Border Approximation Area Comparison (MABAC) methods. The supplier selection process is evaluated using five criteria, namely price, product quality, delivery punctuality, legality and certification, and warranty. Data were collected through observation, interviews, and company documentation. The study used 60 suppliers selected through purposive sampling. SAW was applied to normalize the supplier data, calculate weighted preference values, and select 20 suppliers with the highest preference values for further evaluation. MABAC was then applied to the 20 selected suppliers by calculating their relative positions with respect to the border approximation area. The final ranking shows that PT. Nusa Medika Utama was ranked first with a MABAC value of 0.235408, followed by PT. Global Medika Nusantara and PT. Nusa Sehat Husada, both with a MABAC value of 0.185408. The results provide a ranked list of supplier alternatives that can be used as supporting information for supplier selection at PT. Azkha Medical Indonesia.
Performance Comparison of Bi-LSTM and GRU for Web-Based Stock Price Prediction: A Case Study of BBCA JK Khaerul Oktafiansyah; Muhammad Feizal
Journal of Computer Science and Informatics Engineering Vol 5 No 4 (2026)
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55537/cosie.v5i4.1814

Abstract

Predicting stock prices is complex due to the non-linear and volatile characteristics of financial time-series data. This study presents a comparative analysis and web-based implementation of Bidirectional Long Short-Term Memory (Bi-LSTM) and Gated Recurrent Unit (GRU) architectures to predict the daily closing prices of PT Bank Central Asia Tbk (BBCA.JK). Using Yahoo Finance historical data from January 1, 2018, to December 31, 2024 (1,725 trading days) via a univariate approach, the dataset was partitioned into 70% training, 15% validation, and 15% testing. Empirical results demonstrate that the GRU model significantly outperforms Bi-LSTM, achieving a Root Mean Squared Error (RMSE) of 145.63, a Mean Absolute Error (MAE) of 116.95, and a Mean Absolute Percentage Error (MAPE) of 1.23%. A Paired ttest confirmed the statistical significance of this superiority, yielding a t-statistic of ???? = -4,821 and a pvalue of ???? = 0,00012 ( ???? < 0,05). The GRU model was integrated into a microservices web application using Flask API and Laravel, incorporating a rule-based Decision Support System (DSS) to provide operational trading signals (Buy, Sell, Hold) along with Take Profit and Stop Loss calculations. Evaluation via User Acceptance Testing with 21 respondents yielded a feasibility score of 82.43% ("Highly Feasible")
IoT-Based Aquarium Water Quality Monitoring System Using Mamdani Fuzzy Logic Method Saifullah Hidayat; Agung Siswopranoto
Journal of Computer Science and Informatics Engineering Vol 5 No 4 (2026)
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55537/cosie.v5i4.1816

Abstract

Water quality is a crucial factor affecting the health and survival of freshwater ornamental fish. Manual monitoring is often ineffective because changes in water conditions can occur rapidly and numerical sensor readings are difficult to interpret directly. This study aims to implement the Mamdani Fuzzy Logic method in an Internet of Things (IoT)-based aquarium water quality monitoring system using Total Dissolved Solids (TDS) and temperature sensors. Sensor data are processed through fuzzification, rule-based inference, and defuzzification to classify water quality into three categories: poor, normal, and good. The monitoring results are then displayed in real time through a web-based dashboard. Experimental results indicate that the TDS sensor achieved an average error of 3%, while the temperature sensor achieved an average error of 2.6%. Furthermore, the implementation of the Mamdani Fuzzy Logic method produced an average error of only 0.6% compared with MATLAB simulation results. These findings demonstrate that the proposed system is capable of monitoring aquarium water quality accurately and providing clear decision support for users in maintaining optimal water conditions. The main contribution of this study is the integration of the Mamdani Fuzzy method into an IoT-based monitoring system, implemented directly on the ESP32 microcontroller for real-time inference, unlike previous studies that typically process fuzzy data on servers or cloud platforms. This approach achieves accuracy comparable to MATLAB simulation with an average error of only 0.6%.
Implementation of Priority Scheduling Algorithm in Rajeg Police Patrol Scheduling Web Application Aisyah Aisyah
Journal of Computer Science and Informatics Engineering Vol 5 No 4 (2026)
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55537/cosie.v5i4.1822

Abstract

The Priority Scheduling Algorithm is proposed as the main approach to address patrol scheduling at the Rajeg Police Station, which is still done manually via WhatsApp messages without any systematic prioritization of high-risk areas and times, resulting in various challenges such as difficulties in determining area priorities, delays in information dissemination, and non-integrated documentation. This study aims to design and develop a web-based patrol scheduling application that implements this algorithm, which calculates priority scores based on an area’s vulnerability level and risk level during high-risk hours, then automatically generates a schedule from highest to lowest score and distributes officer assignments evenly through a round-robin load-balancing mechanism. The research method used was qualitative descriptive research through literature review, observation, interviews, needs analysis, design, implementation, and system testing. The algorithm testing results showed a 100% match between the scores generated by the system and manual calculations, while black-box testing of 11 functional scenarios showed a 100% success rate with no functional failures. These results prove that the developed system is capable of improving the accuracy, efficiency, and ease of patrol scheduling and management within the jurisdiction of the Rajeg Police Station
Implementation of the Random Forest and Naive Bayes Algorithms in Predicting Network Outages at the Kominfo South Tangerang Raafa Syahidul Haq Irsi; Indra Kristianto
Journal of Computer Science and Informatics Engineering Vol 5 No 4 (2026)
Publisher : Ali Institute of Research and Publication

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

The operational activities of the South Tangerang City Communication and Information Office (Diskominfo) rely heavily on a stable network infrastructure, making network disruptions a critical issue that directly affects service quality and employee productivity. The existing network monitoring system remains reactive and does not yet incorporate predictive analytics based on historical network capturing data. This study proposes a web-based network disruption prediction model using the Cross Industry Standard Process for Data Mining (CRISP-DM) framework by integrating a severity score and K-Means-based labeling to generate data-driven class labels before applying the Random Forest and Naïve Bayes classification algorithms. The dataset consists of 5,819 records collected from actual network traffic capturing within the operational environment of the South Tangerang City Communication and Information Office, using four primary network parameters: bandwidth, latency, packet loss, and uptime. After the data preparation stage, the Random Forest model achieved an accuracy of 95.78%, precision of 94.74%, recall of 97.63%, and an F1-score of 96.16%. In comparison, the Naïve Bayes model achieved an accuracy of 94.50%, precision of 92.88%, recall of 97.29%, and an F1-score of 95.03%. Feature Importance analysis identified packet loss as the most influential predictor, contributing 82.32% to the classification decision. Based on the overall evaluation, Random Forest demonstrated superior predictive performance and was deployed as the primary model in a Flask-based web application to support proactive network monitoring and network condition analysis.