cover
Contact Name
Muhammad Khoiruddin Harahap
Contact Email
choir.harahap@yahoo.com
Phone
+6282251583783
Journal Mail Official
publikasi@itscience.org
Editorial Address
Medan
Location
Unknown,
Unknown
INDONESIA
Brilliance: Research of Artificial Intelligence
ISSN : -     EISSN : 28079035     DOI : https://doi.org/10.47709
Core Subject : Science, Education,
Brilliance: Research of Artificial Intelligence is The Scientific Journal. Brilliance is published twice in one year, namely in February, May and November. Brilliance aims to promote research in the field of Informatics Engineering which focuses on publishing quality papers about the latest information about Artificial Intelligence. Submitted papers will be reviewed by the Journal and Association technical committee. All articles submitted must be original reports, previously published research results, experimental or theoretical, and colleagues will review. Articles sent to the Brilliance may not be published elsewhere. The manuscript must follow the author guidelines provided by Brilliance and must be reviewed and edited. Brilliance is published by Information Technology and Science (ITScience), a Research Institute in Medan, North Sumatra, Indonesia.
Articles 544 Documents
AI Decision Support for Demand Forecasting and Retail Stock Using Random Forest Zulfia, Anni; Ilfa, Tasya Nadhira; Damia, Zayyani; Sukiman, T. Sukma Achriadi; Karima, Annisa
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.5901

Abstract

Out-of-stock or excess inventory is a major challenge in retail supply chain management, especially in dynamic urban areas. This stock imbalance not only causes financial losses, but can also reduce customer satisfaction due to products being unavailable when needed. This study developed an artificial intelligence (AI)-based decision support system using the Random Forest algorithm to predict daily demand in retail stores. The model was trained using historical sales data that included various variables such as date, product category, and previous sales trends. After the training process, the model was implemented in the form of an interactive web application using Streamlit, which allows users to easily access the system through a browser without the need for special installation. Testing results show that the model is capable of predicting demand for the next 7 days with a fairly good level of accuracy, as indicated by a Mean Absolute Error (MAE) value of ±4.613 units per day. This application not only provides demand predictions but also presents data visualizations and automatic restocking recommendations based on the prediction results. Thus, this system is expected to help store managers make more accurate, efficient, and data-driven restocking decisions. Additionally, the use of Streamlit simplifies the process of distributing the system widely and enhances accessibility for end-users, including those without a technical background. This research opens opportunities for further development through the integration of real-time data and other AI methods to improve prediction accuracy in the future.
The Use of Photodiode Sensors to Detect Sugar Levels in the Human Body Rizky, Muharratul Mina; Ginting, Depi; Sukiman, T Sukma Achriadi
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.6318

Abstract

Diabetes mellitus is a chronic metabolic disorder characterized by elevated blood glucose levels due to impaired insulin production or utilization. Regular monitoring of blood glucose is essential to prevent long-term complications such as neuropathy, nephropathy, retinopathy, and cardiovascular disease. However, conventional finger-prick glucometer methods, while accurate, are invasive, cause discomfort, and often discourage patients from performing frequent checks. To address this limitation, this study presents the design, implementation, and evaluation of a non-invasive glucose monitoring system utilizing a photodiode sensor in conjunction with a near-infrared (NIR) light source operating at wavelengths of 1600–1700 nm. The system architecture comprises an NIR LED as the light emitter, a photodiode as the optical receiver, an Arduino Nano microcontroller for data acquisition and signal processing, and an OLED display for real-time result presentation. During measurement, the user’s fingertip is placed between the LED and photodiode, allowing light to pass through the tissue. Variations in glucose concentration affect the absorption and scattering of NIR light, altering the intensity received by the photodiode. This analog voltage output is digitized using the Arduino’s ADC and converted into glucose levels through a calibration curve derived from reference readings taken using a commercial glucometer. Experimental evaluation was conducted on five human subjects under two physiological conditions—before meals (preprandial) and after meals (postprandial). Each condition was measured three times to minimize variability caused by movement or environmental light interference. The photodiode sensor readings were compared against glucometer results to assess accuracy. The system achieved an average accuracy of 87.1%, with individual measurements ranging from 79.2% to 96.9% before meals and 88.9% to 98.2% after meals. Statistical analysis revealed a mean absolute error (MAE) of 9.83 mg/dL and a correlation coefficient (R²) of 0.934, indicating a strong linear relationship between the two measurement methods. Notably, the system tended to slightly overestimate glucose levels before meals and underestimate them after meals, which may be attributed to physiological variations and optical path differences. The results demonstrate that the proposed photodiode-based NIR sensing system is a promising, low-cost, and user-friendly alternative to conventional invasive glucose monitoring. With further improvements in calibration algorithms, sensor placement stability, and ambient light shielding, this approach has the potential to be integrated into wearable devices, enabling continuous glucose tracking and improving patient adherence to self-monitoring routines.
Implementation of Machine Learning Virtual Medical Assistant Using NLP for Stunting and Healthcare Efficiency in Simalungun Damanik, Abdi Rahim; R.H.Zer, P.P.P.A.N.W.Fikrul Ilmi; Zulpani, Rahmat; Batubara, Egi
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.6475

Abstract

Stunting is a serious public health issue that has long-term impacts on children's physical growth and cognitive development. In the village areas of Simalungun Regency, North Sumatra Province, there are still significant limitations in access to effective health information and services. Low public awareness and a shortage of medical personnel are the main factors contributing to the suboptimal handling of stunting cases.This study aims to develop and implement a Machine Learning model based on Natural Language Processing (NLP) as a Virtual Medical Assistant to support the processes of education, early diagnosis, and health consultation related to stunting. The model is designed to understand user complaints, provide automated responses, and deliver appropriate nutritional recommendations and preventive actions.Training data were collected through interviews with the Simalungun Health Department and consultations with pediatricians, which were then used to build an NLP model focused on classifying stunting risk. Testing results for risk classification using the Random Forest algorithm with the Persen_Sangat_Pendek feature yielded an accuracy of 99%, precision of 99%, recall of 99%, and F1-score of 99%, indicating that the model is highly effective in distinguishing stunting categories. The developed Virtual Medical Assistant application also successfully responded to common public inquiries using NLP-based approaches. This research is expected to make a meaningful contribution to technology-based health services, particularly in rural areas, and serve as a model for developing similar systems in other regions facing comparable conditions.
Expert System for Diagnosing Phone Damage with Repair Shop Recommendations Using CF Method As’saidy, Willyadam Saad; Yasir, Fajar Novriansyah; Kriswinarso, Tri Bondan
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.6648

Abstract

Based on observations conducted by the author through interviews at one of the repair shops in Palopo City, the owner—Zul, in 2024—stated that repair shops tend to direct mobile phone users to other shops when the type of device damage cannot be handled by their own repair shop. This reflects a lack of a centralized diagnostic system that can accurately identify damage and recommend the appropriate service provider. In response to this issue, this study aims to develop an Android-based expert system application for diagnosing mobile phone damage, equipped with a responsive and location-aware repair shop recommendation feature using the Certainty Factor (CF) method. The application was built using the 4D development model (Define, Design, Develop, Disseminate) and utilizes the results of observations, literature reviews, and interviews with local technicians to form its knowledge base and rule sets. The diagnosis process is carried out by calculating the confidence value between selected symptoms and corresponding damage types using a combined CF formula, such as CFcombine = CF1 + CF2 × (1 – CF1). This allows the system to measure the degree of certainty with which a particular diagnosis can be made. User testing involving target users showed a high level of feasibility and satisfaction, with a System Usability Scale (SUS) score of 82.5, falling into the “Acceptable” and “Good” categories. The application has proven effective in identifying various types of device damage and providing accurate, real-time repair shop recommendations based on both user location and the type of damage detected. This offers a relevant and practical digital solution for the community in Palopo City.
Intelligent News Aggregation System with Automatic Classification, Clustering, and Summarization Zulfikar, Ihsan Ghozi; Wibisono, Yudi; Wahyudin, Asep
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.6712

Abstract

The rapid growth of online news content has made it increasingly difficult for users to access relevant information efficiently. This study presents the development of an intelligent web-based news aggregation system that performs automatic classification, clustering, and summarization of Indonesian-language news articles. The system aims to enhance the news reading experience by organizing articles by category and topic, and by providing concise summaries. The system was built using the ADDIE development model, with each AI component trained and evaluated separately. News classification is handled by a BLSTM-2DCNN model trained on the IndoSum dataset, achieving 86% accuracy and an F1-score of 0.85. This model was also applied to classify 37,187 real-world articles scraped from Kompas and TribunNews during June 2025. Topic clustering is performed using K-means with entropy-weighted Bag-of-Words features over 5-day sliding windows. The clustering quality, evaluated using the Calinski-Harabasz Index, ranged from 5.21 to 525.44 with an average of 80.53, indicating varying cluster cohesion. For summarization, a fine-tuned BART model was used to summarize the article closest to each cluster’s centroid. The model achieved ROUGE scores of 0.6389 (ROUGE-1), 0.5458 (ROUGE-2), and 0.6017 (ROUGE-L). The integrated system automatically scrapes news, classifies and clusters articles, and displays generated summaries through a user-friendly web interface. The results show that combining deep learning and natural language processing offers an effective approach for intelligent news aggregation, helping users consume news faster and more meaningfully.
Development of an Academic Services Chatbot Based on Retrieval-Augmented Generation (RAG) Husain, Mohammad Labib; Wibisono, Yudi; Anisyah, Ani
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.6719

Abstract

Higher education institutions struggle to provide accurate and accessible academic information. Traditional chatbots are often limited in capability, while standard Large Language Models (LLMs) pose a significant risk of factual "hallucinations," rendering them unsuitable for official university use where trustworthiness is paramount. This study aims to increase the accessibility and effectiveness of academic services by developing a trustworthy chatbot. The primary objective is to implement the Retrieval-Augmented Generation framework to create a reliable AI assistant that is factually grounded in a verified, domain-specific knowledge base. A knowledge base was constructed from official FPMIPA UPI documents and structured using hierarchical chunking. The system employs a multi-stage RAG pipeline featuring query contextualization and reranking before generation with Gemini 2.5 Pro. Performance was evaluated using metrics from the RAGAS framework on a 100-question dataset categorized into factual, reasoning, and out-of-context queries. The evaluation revealed strong performance on factual queries, achieving a Faithfulness score of 0.9100. A significant performance decrease was observed for reasoning tasks, with Context Recall dropping to 0.5926. Crucially, the system successfully handled 81.25% of out-of-context questions by correctly refusing to answer, thereby effectively preventing hallucination. The RAG framework provides a viable and empirically-validated blueprint for creating a trustworthy conversational AI for higher education. The model proves to be an effective tool for factual information delivery and has strong potential to modernize how student support and academic services are delivered.
Stock Price Prediction Using the ETSFormer Model Case Study: PTBA Atqiya, Muhammad Azka; Riza, Lala Septem; Anisyah, Ani
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.6729

Abstract

The capital market in Indonesia is currently experiencing very rapid development. This growth is significantly evidenced by the increasing number of investors, especially from the millennial and Gen Z demographics. However, this growing investor base also faces a major challenge: high stock price volatility. These fluctuations are triggered by various factors, ranging from domestic economic policies and global geopolitical conditions to rapidly changing market sentiment. This research aims to build a stock price prediction model for PT Bukit Asam Tbk (PTBA) using the ETSFormer architecture, a modern Transformer-based method designed for time-series data. The historical stock price data used in this study covers a five-year period from 2020 to 2025. To ensure optimal model performance, the best model was identified using the Grid Search technique to find the most effective combination of hyperparameters. The results of this study determined that the best model was achieved with the hyperparameters model dimension = 16, batch size = 16, and a learning rate = 0.01, which yielded a validation loss of 0.0074. In the evaluation phase, this model demonstrated solid performance with a MAPE score of 3.28%, an MAE of 86.76, and an RMSE of 117.2. Although the resulting model is quite good at reading long-term trend directions, observations indicate limitations in capturing short-term price volatility. This implies that the model is more suitable for strategic trend analysis than for predicting daily fluctuations.
Designing a 3D Animation-Based Promotional Video for Sublimation Fabric Product of Asietex Silva, Maya; Hamdan, Hamdan; Sunardi, Sunardi
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.6734

Abstract

PT. Asietex Sinar Indopratama is a vertically integrated textile manufacturer specializing in the full production cycle—from spinning yarn to producing finished fabrics. The majority (85%) of its top-line fabric output is composed of cotton or rayon, while the remaining portion utilizes polyester materials. In line with its commitment to ongoing innovation, the company has launched a high-performance Direct-to-Garment (DTG) and sublimation printing division. This division offers enhanced print clarity, vivid color expression, and superior resistance to repeated washing. Such technological advancements have enabled PT. Asietex to respond efficiently and flexibly to market needs in both large-scale and custom textile applications, particularly in the fashion and home décor sectors.The primary objective of this study is to explore how 3D visual presentation techniques influence consumer interest and sales performance for the company’s textile products. The research method included designing 3D fabric product models using Blender 3D software and creating an animated promotional video that showcases key features and advantages of the sublimated fabric line. This promotional content was distributed through various online marketing channels.To assess the impact of the visualization, feedback was collected from a group of potential customers who viewed the animation, and comparative sales data were analyzed before and after the campaign. Survey findings indicate that 85% of participants considered the video to be highly descriptive and useful in understanding the product. Moreover, 78% expressed heightened curiosity and interest in the sublimation fabric after viewing the video.
Implementation of Data Mining for Analyzing Consumer Purchasing Patterns at TeTa Ino Cafe Tjia, Theresia Elvita; Yasir, Fajar Novriansyah; Ekawati, Shindy
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.6767

Abstract

TeTa Ino is a micro, small, and medium enterprise by university students focusing on the production and marketing of innovative products based on butterfly pea tea. The significant decrease in sales volume from approximately 50–80 units per promotional event to only 20–40 units indicates a potential issue in the current marketing strategy. Therefore, this study aims to identify consumer purchasing patterns that can serve as the foundation for developing more targeted marketing strategies to enhance the competitiveness of TeTa Ino. This research employs the Cross Industry Standard Process for Data Mining (CRISP-DM) approach by applying the K-Means Clustering algorithm to four months of transaction data, including variables such as number of transactions, total transaction value, and discounts offered. The analysis resulted in four distinct consumer clusters: passive consumers, loyal consumers, non-loyal consumers, and consumers with moderate purchasing frequency. Each cluster is recommended to be approached with tailored marketing strategies, such as loyalty programs, product benefit education, and bundling promotions. The clustering evaluation achieved a Silhouette Score of 0.9008 and a Calinski Harabasz Score of 7630.34, indicating good segmentation quality and clear separation among clusters. This study concludes that applying the K-Means Clustering algorithm is effective in mapping consumer purchasing behavior as a basis for data-driven marketing strategy formulation. Future research is recommended to incorporate time-related variables and explore other clustering methods to further strengthen the analysis.
Facial Recognition Software for Employee Presence Using Convolutional Neural Network with InceptionV3 Architecture Nicholas, Nicholas; Al Rivan, Muhammad Ezar
Brilliance: Research of Artificial Intelligence Vol. 5 No. 2 (2025): Brilliance: Research of Artificial Intelligence, Article Research November 2025
Publisher : Yayasan Cita Cendekiawan Al Khwarizmi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47709/brilliance.v5i2.6769

Abstract

Presence is a crucial aspect of human resource management that involves recording and monitoring employee attendance. It serves not only for tracking presence but also as the foundation for salary calculation, performance evaluation, and strategic decision-making. While many companies still adopt manual presence systems due to their simplicity, such methods are inefficient, prone to human error, and burdensome in administrative tasks, especially in the presence of growing operational complexity. Moreover, even digital systems like fingerprint scanners are often inflexible, as they require physical presence at designated devices, making them unsuitable for remote or mobile employees. This research developed an Android-based presence application utilizing facial recognition technology with the Convolutional Neural Network method using the InceptionV3 architecture. The system is designed to enable automatic, flexible, and accurate attendance recording both inside and outside the workplace. A website-based system has also been developed for centralized attendance data management. Implementation results show that the Android-based application successfully enables employees to perform attendance both inside and outside the office using facial recognition technology, eliminating the need for manual documentation. Additionally, the web-based system can automatically record and summarize attendance data, simplifying recapitulation processes and reducing administrative workload. The facial recognition model, trained using gradual transfer learning, achieved an accuracy of 97.86% and F1-Score of 97.55%. This application has significant potential to improve the efficiency and flexibility of corporate attendance systems.