cover
Contact Name
Sopiyan Dalis
Contact Email
sopiyan.spd@bsi.ac.id
Phone
+6281380852868
Journal Mail Official
jurnal.paradigma@bsi.a.cid
Editorial Address
Jl. Kramat Raya No.98, Kwitang, Kec. Senen, Kota Jakarta Pusat, DKI Jakarta 10450
Location
Kota adm. jakarta barat,
Dki jakarta
INDONESIA
Paradigma
ISSN : 14105063     EISSN : 25793500     DOI : http://dx.doi.org/10.31294/paradigma
Core Subject : Science,
The Paradigma Journal is intended as a medium for scientific studies of research, thought and analysis-critical issues on Computer Science, Information Systems, and Information Technology, both nationally and internationally. The scientific article refers to theoretical reviews and empirical studies of related sciences, which can be accounted for and disseminated nationally and internationally. Paradigma Journal accepts scientific articles research at Expert Systems, Information Systems, Web Programming, Mobile Programming, Games Programming, Data Mining, and Decision Support Systems.
Articles 90 Documents
Classification of Vegetable Types Using the Convolutional Neural Network (CNN) Algorithm Karunia, Wilis Arum; Puspita, Safriya Murni; Rolliawati, Dwi; Yusuf, Ahmad
Paradigma - Jurnal Komputer dan Informatika Vol. 27 No. 1 (2025): March 2025 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/p.v27i1.7577

Abstract

This study aims to classify vegetable types using the Convolutional Neural Network (CNN) algorithm with a dataset encompassing 15 vegetable classes and a total of 31,000 images. By utilizing the TensorFlow and Keras libraries, the CNN model was designed with convolutional, pooling, and dense layers to recognize visual features such as color, texture, and shape. The results indicate a highest validation accuracy of 95.83% and a testing accuracy of 93%. This research contributes to the application of the CNN algorithm for image classification and demonstrates its potential in handling multi-class datasets effectively. However, since the vegetables used have very distinct shapes and textures, this study is more relevant in the context of the technical application of the CNN algorithm rather than practical benefits. The research would be more impactful if applied to vegetables with similar shapes and characteristics, thereby supporting farmers or individuals studying vegetable traits in greater depth. Additionally, such an approach could address challenges in differentiating visually similar vegetable types, making the technology more valuable in real-world agricultural or educational settings.
Efficient Image Transmission for Autonomous Systems Using Residual Dense Feature Networks Over LoRa Networks Praptawilaga, Muhamad Fadly Rizqy; Suranegara, Galura Muhammad; Satyawan, Arief Suryadi
Paradigma - Jurnal Komputer dan Informatika Vol. 27 No. 1 (2025): March 2025 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/p.v27i1.7584

Abstract

Autonomous systems face challenges in transmitting high-quality images over bandwidth-constrained networks like LoRa, which operates at data rates of 0.3–50 kbps. This study proposes the Residual Dense Feature Network (RDF Net), a super-resolution model designed to optimize image transmission within the constraints of LoRa networks. By leveraging Contrast-Aware Channel Attention (CCA), Enhanced Spatial Attention (ESA), Blueprint Separable Convolution (BSConv), and a progressive approach, RDF Net achieves 20x upscaling, enabling low-resolution images (40x40 pixels) to be reconstructed into high-resolution outputs (800x800 pixels) on a central server. Experimental evaluations demonstrate that Model-4, combining CCA and ESA, delivers state-of-the-art perceptual quality and structural fidelity, while Model-3, using ESA, offers a computationally efficient alternative for resource-constrained scenarios. Simulations of LoRa’s bandwidth limitations reveal that transmitting a single 40x40 image requires approximately 0.208–0.56 seconds at a data rate of 50 kbps. While this demonstrates the feasibility of near real-time communication, the trade-off between latency and visual fidelity remains a critical consideration, particularly for latency-sensitive applications. These findings underscore RDF Net’s potential to address the challenges of high-quality visual communication in bandwidth-constrained environments, paving the way for enhanced autonomous system applications. Further optimization, including adaptive compression strategies, and testing on actual LoRa hardware are recommended to validate its performance in real-world scenarios and explore its applicability to diverse autonomous systems.
Prediction Customer Loyalty Using Random Forest Algorithm on Shopee Reviews Saputra, Ferdi; Fersellia, Fersellia
Paradigma - Jurnal Komputer dan Informatika Vol. 27 No. 1 (2025): March 2025 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/p.v27i1.7940

Abstract

This research develops a Shopee customer loyalty prediction model using Random Forest algorithm, utilizing customer reviews from Google Play Store. One of the key issues in e-commerce is maintaining customer loyalty amidst intense competition, so it is important to identify loyal customers and understand the factors that influence their commitment. This study involves data collection through web scraping, data cleaning, loyalty labeling, and Random Forest-based prediction model building and evaluation. The evaluation process was conducted using a confusion matrix to measure accuracy, precision, recall, and F1-score. The model classified customers into loyal, neutral, and disloyal categories, with an overall accuracy of 97%. The model showed precision, recall, and F1-score of 0.98 for loyal customers, and 0.99, 1.00, and 0.99 for disloyal customers. However, identification of neutral customers is still a challenge, with precision, recall, and F1-score of 0.92, 0.85, and 0.88, respectively. The results of this study provide strategic insights for Shopee in improving customer retention strategies and demonstrate the effectiveness of the Random Forest algorithm in analyzing review data.
Optimizing Employee Admission Selection Using G2M Weighting and MOORA Method Rahmanto, Yuri; Wang, Junhai; Setiawansyah, Setiawansyah; Yudhistira, Aditia; Darwis, Dedi; Suryono, Ryan Randy
Paradigma - Jurnal Komputer dan Informatika Vol. 27 No. 1 (2025): March 2025 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/p.v27i1.8224

Abstract

An objective and effective employee admission selection process is a crucial step for the success of the organization in achieving its goals. Problems in employee recruitment selection often arise due to a lack of good planning and system implementation, namely decisions are often influenced by personal preferences, stereotypes, or non-relevant factors, thus reducing objectivity in choosing the best candidates. Objective selection ensures that candidate assessments are conducted based on measurable, relevant, and bias-free criteria, so that only individuals who truly meet the company's needs and standards are accepted. The purpose of developing an optimal approach in employee admission selection using G2M weighting and MOORA is to create a more objective, efficient, and accurate selection process. This approach aims to integrate the calculation of criterion weights mathematically, such as those offered by G2M, in order to eliminate subjective bias in determining criterion prioritization. The MOORA method of evaluating alternative candidates is carried out through ratio analysis that takes into account various criteria simultaneously, resulting in a transparent and data-driven ranking. The results of the employee admission selection ranking based on the criteria that have been evaluated, Candidate 3 obtained the highest score of 0.4177, indicating that this candidate best meets the expected criteria. The second position was occupied by Candidate 6 with a score of 0.3886, followed by Candidate 9 with a score of 0.3528. This research contributes to the recruitment process, by providing a more reliable, transparent, and less subjective way of selecting the right candidates for the positions that companies need.
Combination of Logarithmic Least Square Weighting and MAUT Method for Best Employee Selection in Retail Companies Saputra, Aditya; Priandika, Adhie Thyo
Paradigma - Jurnal Komputer dan Informatika Vol. 27 No. 1 (2025): March 2025 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/mf9wad40

Abstract

Selecting the best employees plays a crucial role in enhancing the performance of retail companies. Given that each employee has unique roles, responsibilities, and working conditions, creating a truly fair and consistent assessment standard can be challenging. Additionally, subjective factors such as personal bias or preferences of the assessor can influence the evaluation outcome. The integration of LLSW and the MAUT method in employee selection offers a systematic approach that combines precise weighting with multi-criteria utility analysis. This combination aims to improve the accuracy, objectivity, and transparency of the decision-making process. By utilizing both methods, retail companies can establish a more effective, transparent, and data-driven selection system, ensuring that the best employees are chosen based on rational and fair evaluations. The results of the employee selection process using LLSW and MAUT showed that Employee RS ranked first with the highest score of 0.7485, indicating the strongest qualifications compared to the other candidates. Employee LK and Employee ML ranked second and third with scores of 0.6035 and 0.572, respectively, demonstrating solid performance. These selection outcomes can assist companies in recruiting the most suitable workforce for their operational needs and vision, ultimately leading to improved productivity and service quality in the long run. The main contribution of this research is capable of improving accuracy and fairness in employee performance evaluation. This approach reduces the subjectivity that often occurs in conventional assessment processes in the retail sector, as well as providing a basis for transparent and measurable decision-making.
Support Vector Machine with FastText Word Embedding for Hate Speech Aspect Categorization Mardiana, Aida Milati; Rozi, Imam Fahrur; Arianto, Rudy
Paradigma - Jurnal Komputer dan Informatika Vol. 27 No. 2 (2025): September 2025 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/p.v27i2.5127

Abstract

Freedom of expression on Twitter often leads to issues such as hate speech, which may include provocation, incitement, or insults based on race, religion, gender, and other aspects. To address this issue, machine learning techniques can be applied to automatically classify hate speech. Therefore, this study aims to implement a machine learning–based approach for automatic hate speech aspect classification and to evaluate the accuracy of the obtained results. Support Vector Machine is used as the classifier method, with FastText as the word embedding method in the categorization process of hate speech aspects. The categorized aspects include abusive, individual, group, religion, race, physical, gender and other. The dataset used in this research is a collection of Indonesian tweets from Kaggle, which have been classified into each aspect. This study also tested combinations of preprocessing methods, namely filtering with stemming and the FastText pre-trained model. From the test results of the application of the Support Vector Machine method with FastText word embedding, with parameters C value = 1.0, gamma = 1.0 and RBF kernel and the ratio between training data and testing data is 90:10, the best results were obtained accuracy 98%, precision 98%, recall 98% and F1-Score 97% on Physical and Gender aspects. In addition, this study also tested if it did not use fasttext word embedding and the accuracy results showed 84%, precision 74%, recall 86% and F1 Score 79% in the abbusive aspect.
An Objective Sales Team Performance Assessment Model: Integrating Entropy Weighting and Multi-Attribute Utility Theory Efendi, Rio; Thyo Priandika, Adhie
Paradigma - Jurnal Komputer dan Informatika Vol. 28 No. 1 (2026): March 2026 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/p.v28i1.8713

Abstract

Evaluation of sales team performance is essential for measuring the effectiveness of marketing strategies and achieving corporate goals. The main problem in evaluating sales team performance lies in the high subjectivity of assessments, unclear performance indicators, and difficulties in determining objective and consistent criteria weights. This situation results in evaluation outcomes that are unstable and potentially lead to suboptimal managerial decisions. To address this, the study applies a hybrid approach combining Entropy weighting and Multi-Attribute Utility Theory (MAUT). Entropy objectively derives criterion weights from data variability, while MAUT systematically transforms performance scores into utility values, enabling more consistent comparisons than traditional assessment methods. The research results show that the proposed model produces stable and quantitatively consistent rankings, with a utility score range between 0 and 1.0048 reflecting measurable performance differentiation among teams. Team G achieved the highest score of 1.0048, while Team D scored 0, indicating a significant performance gap. Compared to conventional methods, which tend to yield more homogeneous values, this hybrid approach is more effective in minimizing bias, enhancing discriminative power, and strengthening the reliability of managerial decision-making. This research makes a significant contribution to the development of scientific knowledge by presenting an innovation in the form of integrating the Entropy and MAUT methods in the context of sales team performance evaluation that is more objective, systematic, and data-based.
Interactive Live Streaming Website with Cloud Computing and AI Assistant Integration Tantra Adlillah, Satria; Rosita, Ai
Paradigma - Jurnal Komputer dan Informatika Vol. 28 No. 1 (2026): March 2026 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/p.v28i1.9724

Abstract

Live streaming platforms continue to face persistent challenges related to high latency, limited scalability, and insufficient real-time interactivity, which degrade user experience and hinder audience engagement. This study aims to address these limitations by developing StreamCats, a third-party interactive web-based live streaming system that integrates cloud computing infrastructure, WebSocket-based real-time communication, and an AI assistant. The system was developed using Object-Oriented Methodology (OOM) and the Rational Unified Process (RUP), implemented with Next.js, Socket.IO, Supabase, and Puter.js, while TikTok live event connectivity was achieved through WebSocket reverse engineering. System performance was evaluated through stress testing using Apache JMeter under 10, 50, and 100 concurrent user scenarios, functional validation of six core modules, and usability testing involving ten participants. The results demonstrate that under 100 concurrent users, the system maintained an average latency of 2,180.3 ms (well below the 5,000 ms threshold), a response time of 139 ms, CPU usage of 1.13%, and a throughput of 89.8 requests per second, outperforming comparable platforms Indofinity (57 req/s) and Tikfinity (19.5 req/s). All six modules passed functional tests, and 90% of participants successfully completed usability tasks with an average comprehension time of under five minutes. These findings confirm that the integration of cloud computing and AI-driven assistance significantly enhances scalability, responsiveness, and user experience in interactive live streaming environments, establishing StreamCats as a viable and competitive foundation for multi-platform real-time streaming solutions
Comparative Analysis of Multi-Classifier Models with Resampling Techniques for Imbalanced Student Graduation Prediction Carolina, Irmawati; Lia Andharsaputri, Resti; Suharjanti, Suharjanti; Prihatin, Titin; Nurdin, Hafis
Paradigma - Jurnal Komputer dan Informatika Vol. 28 No. 1 (2026): March 2026 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/p.v28i1.11976

Abstract

Student graduation prediction supports early academic intervention but commonly suffers from class imbalance, where on-time graduates dominate the dataset. This study evaluates five classifiers—Random Forest (RF), XGBoost, Logistic Regression (LR), k-Nearest Neighbors (k-NN), and Gaussian Naïve Bayes (GNB)—under five class-imbalance handling scenarios: Baseline (no resampling), Random Undersampling (RUS), SMOTE, ADASYN, and Borderline-SMOTE. Experiments were conducted on 796 student records (10 attributes) with an imbalanced distribution (634 on-time vs. 162 not on-time; ratio 1:3.9) using Stratified 5-Fold Cross-Validation. Performance was assessed using confusion-matrix metrics and AUC-ROC to reflect minority-class detection. Under baseline, RF achieved the highest accuracy (0.873) but limited minority recall (0.573), confirming majority-class bias. Resampling consistently improved minority recall across models; for example, LR recall increased to 0.802 with RUS, while GNB reached 0.833 with ADASYN, although accuracy decreased due to the sensitivity–specificity trade-off. Overall, RF and XGBoost showed the most stable discrimination across resampling scenarios based on AUC (RF: 0.870–0.883; XGBoost: 0.847–0.866). The main contribution is a systematic, reproducible comparative evaluation of classifier–resampling combinations for imbalanced graduation prediction, providing practical guidance for selecting robust models to identify students at risk of delayed graduation.  
Integration of LODECI Weighting Method and SPOTIS in Employee Performance Evaluation Based on Multi-Criteria Decision Making  Shely Amalia, Fadila; Darwis, Dedi; R. Mehta, Abhishek
Paradigma - Jurnal Komputer dan Informatika Vol. 28 No. 1 (2026): March 2026 Period
Publisher : LPPM Universitas Bina Sarana Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31294/p.v28i1.12508

Abstract

Employee performance evaluation in many organizations often faces challenges due to numerous assessment criteria and potential subjectivity in the decision-making process, making the evaluation results less consistent and objective. Multi-Criteria Decision Making (MCDM) methods have been widely used to address this problem; however, previous approaches generally still rely on subjective weight determination and do not fully consider the stability of results against data variation. Therefore, this study aims to develop a more objective and stable decision-making model by integrating the LODECI method to determine criteria weights based on data and the SPOTIS method to rank alternatives based on their distance from the ideal solution. Five evaluation criteria are used, namely productivity, work quality, discipline, teamwork, and responsibility, with data collected from eight employees as alternatives. The analysis process was carried out through the stages of constructing a decision matrix, calculating criterion weights using LODECI, and ranking using SPOTIS which produced a total distance value as a quantitative evaluation metric. The research results show that GS Employee achieved the smallest distance value of 0.058, thus ranking first, followed by CR Employee with a value of 0.086 and AN Employee with a value of 0.321. These findings indicate that the proposed model is capable of providing more measurable and consistent evaluation results. The main contribution of this study lies in the integration of objective weighting and ideal-solution-based ranking methods supported by sensitivity analysis, thereby producing a performance evaluation system that is more reliable, transparent, and robust compared to previous approaches.