Sinkron : jurnal dan penelitian teknik informatika
Not a member yet
1261 research outputs found
Sort by
A Disaster-Aware Traffic Assignment Model: Comparative Evaluation of Frank-Wolfe and Simulated Annealing Algorithms
No full textTraffic assignment under disaster-induced disruptions poses unique challenges, as traditional models often overlook sudden capacity loss and unpredictable demand. This study introduces a disaster-aware Traffic Assignment Problem (TAP) model that integrates a modified Bureau of Public Roads (BPR) cost function, explicitly accounting for effective capacity changes during disasters. The Frank-Wolfe (FW) algorithm is applied to solve the model, chosen for its scalability and convergence properties. A comparative analysis with Simulated Annealing (SA) is also performed across various network sizes and disruption scenarios. Results show that FW consistently delivers near-optimal flow distributions with lower travel costs and faster convergence. While SA exhibits higher variability under tight capacity constraints, FW demonstrates robust stability, particularly in medium to large networks under moderate to severe disruptions. Flow patterns from FW highlight adaptive traffic redistribution, effectively bypassing congested or blocked links. This study is the first to systematically compare Frank-Wolfe and Simulated Annealing under disaster-induced TAP conditions with capacity degradation. Contributions include (1) formulating a disaster-aware TAP model, (2) applying FW to disrupted networks, and (3) validating through structured simulations. Findings suggest that FW offers a reliable optimization tool for real-time traffic reallocation, supporting resilient urban mobility in emergencies
Hybrid CNN and KNN Approach for Coffee Bean Quality Identification
No full textThis study discusses the integration of Convolutional Neural Network (CNN) and K-Nearest Neighbors (KNN) for the identification of coffee bean quality as an effort to increase the competitiveness of local commodities. CNN is used as a feature extractor to produce an information-rich representation of coffee bean images, while KNN acts as a classifier to classify quality into two classes, namely Good and Defective. The dataset is divided into training, validation, and test data, with a total of 1,190 images obtained from the manual annotation process. The research stages include (1) pre-processing of data in the form of cropping based on bounding boxes, resize to 224×224 pixels, normalization, and data augmentation; (2) feature extraction using pretrained CNN (ResNet18) by eliminating the final classification layer to obtain a 512-dimensional feature vector; and (3) classification using KNN with variations in k values (3, 5, and 7) as well as Euclidean distance metrics. The results of the experiment showed that the CNN+Softmax baseline resulted in an accuracy of 86%, while the CNN+KNN method provided better performance. The k=5 configuration was proven to be optimal with an accuracy of 93.4%, precision, recall, and a balanced F1-score in both classes. The confusion matrix shows that most samples can be classified correctly with a low error rate. These findings are in line with previous research that emphasized the effectiveness of CNN in the extraction of visual features and the advantages of KNN on limited datasets. Thus, this approach can be a practical solution to support an automatic, accurate, and consistent coffee bean quality identification system to increase the competitiveness of local coffee commodities in the global market
Usability and UX of Ruangguru Mobile: A 120-User Evaluation with SUS and UEQ
No full textThe rapid growth of digital learning in Indonesia has emphasized the importance of evaluating the user interface (UI) and user experience (UX) of mobile learning platforms such as Ruangguru. This study aims to assess the usability and experiential quality of the Ruangguru mobile application to understand how effectively it supports engagement and learning motivation. A quantitative descriptive method was employed using two standardized instruments: the System Usability Scale (SUS) and the User Experience Questionnaire (UEQ). The SUS, consisting of 10 Likert-scale items, measures usability, while the UEQ evaluates six dimensions—Attractiveness, Perspicuity, Efficiency, Dependability, Stimulation, and Novelty—on a seven-point semantic differential scale. Data were collected from 120 purposively selected users (high school and university students who had used the app for at least one month and once a week) through Google Forms. Results show a mean SUS score of 76.2, classified as Good/Acceptable, exceeding the global usability benchmark of 68. UEQ results indicate Excellent ratings for Stimulation (1.89) and Novelty (1.95), and Good ratings for Attractiveness (1.61), Perspicuity (1.42), Efficiency (1.68), and Dependability (1.44). These findings highlight Ruangguru’s strong emotional engagement through gamification and visual design, though minor dependability issues suggest optimization for latency and consistency. Overall, Ruangguru demonstrates an intuitive, motivating, and functionally robust design. Future improvements should prioritize navigation flow, performance stability, and cross-device optimization. This study underscores the value of combining SUS and UEQ as a dual-instrument framework for evidence-based, user-centered design in digital learning environments
Integrating Blockchain with Neural Networks for Forest Fire Classification
No full textForest fires represent one of the most severe environmental disasters, causing extensive ecological, social, and economic damage—particularly in tropical nations like Indonesia. This research introduces a hybrid framework that combines Blockchain and Neural Network technologies to classify forest fire images. The goal is not only to enhance detection precision but also to guarantee the integrity and security of experimental data. Two deep learning architectures, ResNet-50 and VGG-16, were implemented and evaluated to compare their effectiveness in differentiating fire from non-fire imagery. The dataset merges locally collected images from the Puncak area of Bogor, Indonesia, with the public FIRE dataset from Kaggle, thereby increasing model generalization. Model training utilized a transfer learning strategy, and its performance was assessed through four key indicators: accuracy, precision, recall, and F1-score. The findings demonstrate that VGG-16 achieved the most reliable outcomes, obtaining an accuracy of 0.91 and an F1-score of 0.87, outperforming ResNet-50, which reached 0.82 accuracy. All experimental data, including training and inference outputs, were stored using the InterPlanetary File System (IPFS), while each file’s Content Identifier (CID) and metadata were recorded in a blockchain-based smart contract to ensure transparency, verifiability, and data permanence. The study concludes that integrating blockchain with deep learning establishes a trustworthy and tamper-resistant framework for forest fire image classification. Future research may explore lighter CNN models and the fusion of IoT sensor data to enable adaptive and real-time fire detection
Implementation of LSA for Topic Modeling on Tweets with the Keyword ‘Kemenkeu’
No full textThis research explores public discourse on financial policies by analyzing tweets mentioning the keyword 'Kemenkeu' (Ministry of Finance). Using Latent Semantic Analysis (LSA), the study examined 10,099 tweets to uncover key topics that reflect public sentiment toward the Ministry’s policies. Preprocessing steps, such as stopword removal and stemming with Sastrawi, were essential to ensure the effectiveness of the analysis. The results revealed three main topics: Finance and Budget, Salaries and Employee Welfare, and Excise and Customs Regulations. These insights provide a better understanding of public opinion on financial issues and highlight the importance of proper text preprocessing in topic modeling. This approach demonstrates how LSA can be used as a tool for analyzing large-scale social media data, offering valuable input for policymakers. Future research could expand on this by using more advanced models or larger datasets to gain deeper insights
Performance Level Analysis On Learning Vector Quantization And Cohonen Algorithms
No full textBiometric identification is an alternative for a security system that consists of physiological characteristics and behavioral characteristics. Physiological characteristics are relatively stable physical characteristics such as fingerprints, hand lines, facial features, tooth patterns, and the retina of the eye. Behavioral characteristics such as signature, speech patterns, or typing rhythm. The function of a signature is proof in a document which states that the party signing, knows and agrees to all the contents of a document. There are several stages in the signature pattern image recognition system, namely the signature pattern image is produced through a scanning process, then the resulting digital signature image is cut (scaling) manually, the next process is thresholding, edge detection, image division, and representation. input value. The method used in recognizing signature patterns is the learning vector quantization (LVQ) artificial neural network method and kohonen self-organizing map (SOM). In Learning vector quantization, the initial weights are updated using existing patterns. Meanwhile, in the self-organizing map method, Kohonen takes initial weights randomly, then these weights are updated until they can classify themselves into the desired number of classes. The processes that occur in the artificial neural network method require a relatively long time. This is influenced by the large number of data samples used as a means of updating the trained weights. From the results of the research conducted, it shows that the learning rate value that was built around 0.2 < α ≤ (10) ^ (-2) can produce better signature pattern recognition accuracy
Comparative Analysis of Express and Hono Framework Performance in Simple Registration Application
No full textThis research evaluates the performance of two Node.js frameworks, Express and Hono, in developing a simple registration application. This application serves as a backend to store user registration data into a PostgreSQL database using the pg client of the node package manager (npm). The purpose of this performance comparison is to identify the framework that is superior in executing 1 million requests in this scenario. The analysis shows that Express has an average execution time of 26.85% faster than Hono. However, it is inversely proportional to the resource usage, where Hono shows better efficiency with lower CPU and memory usage of 29.29% and 19.97%. These findings provide important insights for developers in choosing a suitable framework based on performance and resource efficiency requirements
Graph Regularized Probabilistic Latent Semantic Analysis for Topic Analysis Using Social Media Data
No full textIn today's digital era, social media data provides valuable insights into public opinion. This study implements the Graph Regularized Probabilistic Latent Semantic Analysis (GPLSA) method to analyze topics from social media data surrounding the 2024 Indonesian Presidential Election (Pemilu), as well as to evaluate the efficiency of the Probabilistic Latent Semantic Analysis (PLSA) algorithm. The research stages include collecting social media data on presidential debates and elections, text pre-processing, and applying the GPLSA method to identify main topics. The analysis results show that PLSA without graph achieved a topic coherence score of 0.653, indicating good consistency, while GPLSA decreased to 0.5, suggesting that the addition of graph regularization did not significantly enhance coherence. Additionally, PLSA without graph achieved a perplexity score of 12.138, indicating good predictive capability, while GPLSA increased to 12.511, showing that graph regularization did not improve the prediction of new words. PLSA without graph also produced topics relevant to election issues, while GPLSA generated topics influenced by graph regularization, though without significant improvement in topic quality. Sentiment analysis of social media posts provides insights into public responses to debates and election issues. Validation of the GPLSA model ensures relevant topic representation. This research contributes to the development of text analysis methods and offers valuable information for elections and democratic participation. These results can be utilized by stakeholders to make more strategic and informed decisions
KNN Approach to Evaluating the Feasibility of Using Scientific Publications as Final Projects
No full textThis study aims to explore the feasibility of using scientific publications as a substitute for traditional final assignments in higher education by applying the K-Nearest Neighbors (K-NN) algorithm. Traditional final assessments, such as theses, are widely used in evaluating students, but with the increasing availability of peer-reviewed scientific publications, there is potential to use them as a more dynamic and relevant assessment tool. This study uses a dataset containing scientific publications and theses, with features such as research quality, relevance, methodology, and clarity. This study applies the K-NN algorithm to classify these materials and determine whether scientific publications can serve as an effective substitute. The results show that the K-NN algorithm, using k=4, achieved 95% accuracy, successfully distinguishing between scientific publications and theses. However, some misclassifications occurred, indicating areas for improvement, such as incorporating additional features such as citation counts or peer-review scores. These findings suggest that scientific publications, if properly classified, can indeed replace traditional final assignments, encouraging critical thinking and engagement with current research. Future research should refine the feature set and explore other machine learning models to improve accuracy. The practical implications of this research are the potential to develop more innovative and relevant approaches to assessment in higher education, which are more aligned with modern educational practice
Implementation of Support Vector Machine Algorithm for Heart Disease Risk Identification Using Signal Electrocardiogram
No full text: In the medical world, one of the biggest contributors to death in the world is heart disease. Early detection of the risk of heart disease can increase the chances of recovery and reduce mortality. This research applies the Support Vector Machine (SVM) algorithm to identify the risk of heart disease using Electrocardiogram Signals. The ECG data used was taken from a public database that contained a record of information on the electrical activity of the heart of patients with various heart health conditions. The Support Vector Machine algorithm is applied to classify ECG signals into 2 main classes, namely normal conditions and at-risk conditions. Several methods in data processing, including data normalization and feature selection are used to improve the accuracy and success of the model. The results of the evaluation with this method resulted in accuracy, precision, recall and also F1-score showed that the modeling of this algorithm produced a fairly good classification, with an accuracy of more than 90% in the identification of heart disease risk. This study shows the potential use of this algorithm in automatically detecting the risk of heart disease based on ECG signals, which can be a tool in medical diagnosis. The results show that implementing the SVM strategi with the RBF kernel appears to be a very easy execution when compared to the direct part. An important component that affects the adequacy of an SVM strategy is the parameters of the section and the way the information is handled