Publication Search

User reviews on the Shopee e-commerce platform represent an important source of information for understanding consumer perceptions of products and services. Sentiment analysis is commonly applied to classify user opinions into positive, neutral, and negative sentiment categories based on textual data. This study aims to analyze the performance of the Long Short-Term Memory (LSTM) method in sentiment classification of Shopee user reviews. The dataset used in this study consists of Indonesian-language user reviews that have undergone preprocessing stages, including case folding, text cleaning, tokenization, and stopword removal. The LSTM model was trained using preprocessed text represented as word sequences. Model performance was evaluated using overall accuracy and class-wise classification results. The experimental results indicate that the LSTM method achieved an overall accuracy of 87.62%. In addition, the classification performance for the positive sentiment class reached 95.27%, the neutral class achieved 4.96%, and the negative class reached 74.26%. These results demonstrate that the LSTM method performs well in classifying sentiment in Shopee user reviews, particularly for positive sentiment. This study is expected to provide insights and references for the application of deep learning methods in sentiment analysis of Indonesian e-commerce review data.

This research aims to develop a multiclass flower image classification system using the Convolutional Neural Network (CNN) algorithm with the EfficientNet architecture. The main problem addressed is the difficulty of manual identification of flower species that share high visual similarity. The research stages include collecting 17,299 flower images across 19 classes, performing data preprocessing such as image resizing, pixel normalization, and augmentation, followed by model training using the EfficientNet transfer learning approach. The model was trained for 10 epochs with an 80:20 training-validation data split. The evaluation results show that the model achieved a validation accuracy of 98.05% with a loss value of 0.0968, and an average precision, recall, and F1-score of 0.98. The trained model was then implemented into a web-based application built using the Next.js framework, enabling users to upload flower images and obtain real-time classification results via the Hugging Face API. The system successfully identified flower species with a confidence level of 99.87%. These findings demonstrate that combining a modern CNN architecture with transfer learning provides efficient and highly accurate flower classification performance, which can be effectively implemented for educational and digital conservation purposes.

Traditional association rule analysis is effective at uncovering co-purchase patterns but fails to provide a global structural view of the market, which often results in fragmented and isolated insights. This study proposes a hybrid framework that integrates the Apriori algorithm with a Minimum Spanning Tree (MST) in order to validate and contextualize association rules within a single structural backbone. Transaction data from a retail store are transformed into a weighted, undirected product graph using an inverse-support function, and an MST is then extracted to represent the market backbone, while frequent itemsets and strong rules are obtained using Apriori. Experimental results on 236 multi-item transactions show that the MST backbone comprises 10 products and 9 fundamental links, with 66.67% of these links being confirmed by strong association rules, indicating a substantial coherence between statistical and structural evidence. The proposed model identifies 41 Apriori patterns that can be embedded in the MST and ranks them using a new metric, Structural Distance, which enables the categorization of Core Patterns, Bridge Patterns, and Complex Patterns according to their structural tightness. This hybrid perspective distinguishes dense, strategically meaningful bundles from anomalous but frequent combinations that are structurally peripheral, thereby offering a more holistic and actionable alternative to conventional Market Basket Analysis. The validated framework can support various applications, including store layout optimization, cross-selling strategies, and the design of path-based recommender systems, and it opens avenues for future extensions based on dynamic graphs and Graph Neural Networks.

Detection and identification of plant diseases is critical to the success and efficiency of agricultural production. Plant disease outbreaks are becoming more frequent throughout the world, and the presence of these diseases in cultivated plants has a significant impact on productivity. Therefore, researchers are focusing on developing effective and reliable plant disease detection methods. Thus, farmers can take advantage of early detection of this disease to minimize future losses. This article discusses machine learning approaches as well as decision trees, K-nearest neighbors, naive Bayes, support vector machines (SVM), and random forests for detecting coffee leaf diseases using leaf images. The above-mentioned classifications were researched and compared to determine the most suitable plant disease prediction model with the highest accuracy. Compared with other classification algorithms, the SVM algorithm achieves the highest accuracy of 99.75%. All the models trained above will be used by farmers to quickly identify and classify new diseases in images as a prevention strategy. As a preventive measure, farmers can detect and classify new diseases in images early.

Perkembangan kecerdasan buatan (AI) mendorong inovasi dalam analisis keuangan, termasuk prediksi harga saham yang fluktuatif. Penelitian ini bertujuan memprediksi harga saham PT Garudafood Putra Putri Jaya Tbk menggunakan model ARIMA dengan penanganan Outlier sebagai pendekatan awal menuju sistem prediksi yang lebih adaptif. Data harga penutupan harian dari Yahoo Finance dianalisis melalui uji stasioneritas, identifikasi model ARIMA, deteksi Outlier berbasis log-return, serta evaluasi performa menggunakan RMSE, MAE, dan MAPE. Hasil penelitian menunjukkan bahwa ARIMA Outlier memberikan performa lebih baik dibandingkan ARIMA dasar. ARIMA standar menghasilkan MAPE 1.32% dan AIC –899.46, sedangkan ARIMA dengan tiga dummy Outlier mencapai MAPE 1.16% dan AIC –900.37. Peramalan 14 hari ke depan menunjukkan pola yang stabil pada kisaran Rp 370–371. Pada data uji, ARIMA dasar memberikan akurasi terbaik pada pertengahan Agustus, sedangkan ARIMA Outlier mencapai akurasi tertinggi pada akhir Agustus dengan prediksi Rp 370.2 yang sangat dekat dengan harga aktual Rp 370.4. Hasil ini menunjukkan bahwa penanganan Outlier meningkatkan ketepatan model, sehingga ARIMA Outlier dapat digunakan sebagai fondasi awal menuju pengembangan sistem prediksi keuangan berbasis AI.

Penelitian ini bertujuan untuk merancang dan merealisasikan antena mikrostrip patch berbentuk segitiga (triangular) yang bekerja pada frekuensi 1800 MHz, sebagai penguat jaringan internet di kawasan Kebun Raya Balikpapan. Lokasi tersebut diketahui memiliki kualitas sinyal yang kurang baik untuk beberapa provider. Antena yang dirancang menggunakan dua konfigurasi, yaitu MIMO 8x8 dengan teknik pencatuan langsung dan array 8x1, yang kemudian disimulasikan menggunakan CST Studio Suite 2019 dan diuji performansinya menggunakan metode Speedtest. Hasil simulasi menunjukkan bahwa antena 8x8 memiliki nilai VSWR sebesar 1,50, return loss -12,96 dB, dan gain 4,06 dBi dengan pola radiasi omnidirectional. Sementara itu, konfigurasi 8x1 array menunjukkan nilai VSWR 1,24, return loss -19,21 dB, dan gain 6,46 dBi, juga dengan pola radiasi omnidirectional. Hasil pengujian di lapangan dengan tiga kondisi tanpa antena eksternal, antena MIMO dan antena array 8x1 didapatkan hasil bahwa antena array 8x1 memberikan kecepatan unduh tertinggi mencapai 17,227 Mbps, unggah 4,072 Mbps, Jitter 34,9 ms  dan packet loss 14,18 % sedangkan antena MIMO 8x8 memberikan kecepatan unduh tertinggi mencapai 13,77 Mbps, unggah 2,462 Mbps, Jitter 49,9 ms  dan packet loss 18,43%. Kesimpulannya, kedua jenis antena eksternal yakni MIMO 8x8 dan array 8x1 mampu meningkatkan performa jaringan secara signifikan dan dapat menjadi solusi efektif untuk daerah dengan sinyal lemah. Antena ini diharapkan dapat memberikan kontribusi terhadap akses internet yang lebih stabil di area public yang sangat luas.

The performance of the Indonesian National Team (Timnas) in the 2026 World Cup qualifications has triggered massive and diverse responses on social media, particularly on platform X. This study aims to identify and classify public sentiment regarding Timnas Indonesia's performance into positive, negative, and neutral categories using a data mining approach. Text data was processed through pre-processing stages, term weighting using TF-IDF, and the application of the Synthetic Minority Over-sampling Technique (SMOTE) to address significant class distribution imbalance. The classification algorithm employed was Multinomial Naïve Bayes. Model performance evaluation was conducted by comparing two training-testing data split scenarios: 90:10 and 80:20 ratios. The results indicate that public opinion is dominated by negative sentiment at 73.2%, reflecting public disappointment. In terms of model performance, the 90:10 ratio scenario yielded the best accuracy of 80%, outperforming the 80:20 ratio which recorded an accuracy of 75%. These findings demonstrate that combining Multinomial Naïve Bayes with the SMOTE technique is effective in handling imbalanced text data and is capable of accurately mapping public perception.

The development of social media platforms like TikTok has created new spaces for digital economic activities, including the practive of thrifting, which has now become a trend among the public. However, government policies that block these activities have sparked various public reactions. This study aims to analyze public sentiment regarding the issue of thrifting bans on the TikTok platform using the Bidirectional Long Short-Term Memory (Bi-LSTM) method. This method was chosen because it can understand text context from both directions, allowing it to capture deeper semantic meaning. The dataset consist of 4,000 TikTok user comments collected through a crawling process. The research stages include data preprocessing, sentiment labeling, splitting training and test data, training the Bi-LSTM model, and evaluating performance using accuracy, precision, recall, and F1-score metrics. The research results show that the Bi-LSTM model achieved an accuracy of 86.15%, with stable classification performance and minimal error rate. These findings indicate that Bi-LSTM is effective for sentiment analysis of public opinions on Indonesian language social media, particularly on context specific policy issues. Further development can be carried out by adding pre-trained embeddings or attention mechanisms to improve the model’s performance.

Penelitian ini bertujuan untuk mengembangkan sistem telemetri guna memantau kinerja panel surya pada beberapa lokasi dengan menggunakan komunikasi jarak jauh dan platform Internet of Things (IoT). Metode pemantauan konvensional memiliki keterbatasan dalam menyediakan data secara real-time pada area yang luas, sehingga evaluasi kinerja jarak jauh menjadi kurang efisien. Untuk mengatasi permasalahan tersebut, dirancang sebuah sistem pemantauan menggunakan mikrokontroler ESP32, sensor INA219 untuk mengukur tegangan dan arus, modul GPS Neo-M8 untuk identifikasi lokasi, modul Real-Time Clock (RTC) DS3231 untuk pencatatan waktu, serta modul LoRa RA-02 sebagai media komunikasi nirkabel. Setiap node pengirim dilengkapi dengan modul MicroSD untuk menyimpan data pengukuran secara lokal. Data hasil pengukuran dikirimkan melalui LoRa ke unit penerima dan ditampilkan secara real-time pada platform Thinger.io. Hasil kalibrasi menunjukkan bahwa sensor INA219 memiliki rata-rata galat pengukuran arus sebesar 0,71% dan galat pengukuran tegangan sebesar 0,1%. Pengujian GPS menunjukkan koordinat lokasi yang stabil dengan tingkat akurasi sekitar ±3 hingga ±8 meter. Seluruh data pengukuran berhasil dikirim, disimpan, dan ditampilkan tanpa kehilangan data yang signifikan. Hasil penelitian menunjukkan bahwa sistem yang dikembangkan mampu menyediakan pemantauan parameter panel surya secara jarak jauh yang andal dan efisien dalam kondisi lapangan.

This paper explores the epistemological dimensions of the digital transformation occurring in traditional game development through the integration of machine learning systems. By examining how knowledge creation, validation, and application have evolved in this domain, we identify fundamental shifts in the epistemological frameworks governing game development practices. The research investigates how machine learning has redefined creative processes, technical implementation, and experiential design while challenging traditional notions of authorship, expertise, and knowledge transmission. Through analysis of industry case studies, technological capabilities, and theoretical frameworks, this paper contributes to understanding how machine learning systems are not merely tools but epistemological agents that fundamentally transform how knowledge is generated, validated, and utilized in game development ecosystems.

Manual quality assessment of Platelet Concentrate (TC) is highly subjective and inconsistent, necessitating an objective, automated classification system. This study aims to develop a computationally efficient, low-cost model for TC quality classification using Histogram Features extracted from grayscale images combined with the K-Nearest Neighbor (KNN) algorithm. The methodology employed critical preprocessing steps, including StandardScaler for normalization and SMOTE for balancing the training data, followed by optimization across K=1 to K=30. The optimal model achieved a maximum accuracy of 69.23% at K=6, with an F1-Score of 71.43%, confirming robust performance on the imbalanced testing set. The results validate the effectiveness of the Histogram-KNN approach as a consistent and reliable decision support system for rapid TC quality screening in resource-limited settings.

The development of renewable energy, particularly Solar Power Plants (PV), requires a reliable, real-time, and easily accessible electrical energy monitoring system to ensure optimal system performance. This study aims to design and implement an Internet of Things (IoT)-based electrical energy monitoring system for PV using the NodeMCU ESP32 microcontroller, the PZEM-004T sensor for measuring electrical parameters, and the Node-RED platform as the data visualization interface. The developed system is designed to monitor voltage, current, power, energy, frequency, and power loss in real time, and then display the data in the form of numerical values, graphs, and indicators on a dashboard accessible through a local network. The research method includes hardware design, software development (sensor reading, data processing, and communication), integration with Node-RED, and system testing on a small-scale PV installation. The test results show that the system is capable of monitoring electrical parameters in a stable and responsive manner. Variations in sunlight intensity were found to affect the current and power produced by the solar panels, whereas the inverter output voltage tended to remain within normal operating ranges. The Node-RED dashboard display was considered informative and helpful for users in monitoring and analyzing PV performance. Based on these results, it can be concluded that the IoT-based electrical energy monitoring system designed in this study functions well and is feasible for application in residential or educational-scale PV installations. The system still has the potential for further development through cloud service integration, the addition of environmental sensors, and enhancements to data analysis features and user interface design.

Polusi udara partikulat halus (PM2,5) merupakan ancaman serius bagi kesehatan masyarakat di Kabupaten Brebes, Jawa Tengah. Faktor penyumbang utamanya adalah emisi kendaraan di jalur Pantura, aktivitas industri perikanan, serta konsentrasi tinggi selama musim kemarau (Juni–November). Tidak adanya model peramalan sub-jam yang akurat menghambat pengembangan sistem peringatan dini yang efektif. Penelitian ini mengembangkan dan mengevaluasi model deep learning berbasis Transformer untuk memprediksi konsentrasi PM2,5 dengan resolusi waktu 15 menit. Data yang digunakan berasal dari NASA GEOS-CF (band PM25_RH35_GCC) yang diakses melalui Google Earth Engine menggunakan API Python. Dataset mencakup periode 1 Januari hingga 22 November 2025, menghasilkan 7.813 observasi per jam, yang kemudian diinterpolasi linear menjadi 31.249 titik data dengan resolusi 15 menit. Arsitektur Transformer terdiri dari 3 lapis enkoder, 4 kepala perhatian multi-head, dimensi embedding 128, dimensi feed-forward 256, panjang sekuen 60 timestep, dan augmentasi fitur menggunakan rerata bergulir (*rolling mean*, jendela = 3) dan beda pertama (*first difference*). Pelatihan dilakukan dengan TensorFlow-Keras, pengoptimal Adam, penjadwal peluruhan kosinus (*cosine decay scheduler*), dan fungsi kerugian Huber. Pembagian data dilakukan secara kronologis: 70% pelatihan, 30% validasi. Evaluasi pada set uji independen (16 Agustus–21 November 2025, 9.357 observasi atau 97 hari 11 jam 15 menit) menghasilkan MAE 0,7691 µg/m³, RMSE 1,2052 µg/m³, R² 0,9945, dan *Explained Variance Score* 0,9948. Model ini mampu menggambarkan variasi diurnal dan anomali musiman secara akurat, jauh melampaui model LSTM dan GTWR konvensional. Penelitian ini memberikan kontribusi signifikan di bidang Teknologi Informasi melalui kerangka kerja pengolahan *big data* satelit untuk aplikasi lingkungan.

Tourism is one of the sectors that plays an important role in boosting economic growth through travel activities and destination exploration. Tourists' preferences for nature-based tourism options, such as mountain hiking or beach tourism, are influenced by various factors, ranging from personal experiences and recreational interests to social characteristics. Therefore, a technology-based approach is needed to predict destination choice tendencies more accurately. As artificial intelligence technology develops, deep learning methods have been widely used in classification processes due to their ability to process large amounts of data and recognize complex patterns. In this study, a Multilayer Perceptron (MLP) model is used to classify tourists' preferences between mountain or beach destinations based on a survey dataset. The research stages include data processing, data splitting using a train-test split, model training, and performance evaluation using accuracy, precision, recall, and F1-score. The test results show that the MLP model is capable of achieving an accuracy rate of 99%, confirming that deep learning methods are effective in automatically mapping tourism preference trends. This research is expected to serve as a basis for the development of more personalized travel destination recommendation systems, as well as to support tourism management in formulating targeted promotional strategies.

The increasing global energy demand drives the search for efficient and sustainable renewable energy solutions. Solar panels have become one of the most widely used technologies; however, their efficiency remains limited when installed in a static position. This research aims to analyze the performance of a single-axis auto tracking system on a 10WP solar panel integrated with the Internet of Things (IoT) for real-time monitoring, specifically in powering a portable powerbank. The research method employed was a quantitative experimental design with three testing scenarios: powerbank charging using an auto-tracking solar panel, a static solar panel, and conventional household electricity as a comparison. Charging data were collected via an IoT system integrated with the Blynk application in real-time. The results indicate that the auto-tracking system increased charging efficiency by around 10%, compared to only 6% with a static panel in one hour. This performance is nearly equal to household electricity charging, which reached approximately 10–11%. The study concludes that the single-axis IoT-based auto-tracking system significantly enhances the performance of small-scale solar panels and holds strong potential for portable energy solutions in remote areas.

Suhu dan kelembaban merupakan parameter lingkungan yang harus dijaga pada ruang kubikel untuk memastikan peralatan distribusi listrik tetap bekerja secara optimal. Pada multi-kubikel, perbedaan fungsi dan beban menyebabkan karakteristik suhu dan kelembaban pada tiap ruang kubikel tidak sama, sehingga pemantauan secara manual menjadi kurang efektif dan efisien. Penelitian ini bertujuan untuk merancang dan membangun prototype sistem monitoring dan kontrol suhu-kelembaban pada multi-kubikel berbasis Internet of Things (IoT) yang terdiri dari tiga buah kubikel. Sistem ini menggunakan ESP8266 sebagai mikrokontroler utama dan sensor DHT20 sebagai sensor suhu dan kelembaban yang masing-masing dipasang pada kubikel dengan kondisi lingkungan berbeda. Sistem dilengkapi dengan aktuator kipas dan lampu, serta notifikasi real-time melalui LCD dan Telegram. Meskipun kontrol dan monitoring dilakukan secara terpisah pada tiap kubikel, notifikasi kondisi seluruh kubikel terintegrasi pada satu kanal Telegram yang sama. Pengujian kinerja sistem dengan memberikan variasi suhu dan kelembaban yang berbeda untuk tiap kubikel. Kubikel 1 diberi kondisi normal (suhu 35°C-40°C dan kelembaban 50%-70%), kubikel 2 diberi kondisi overheat (suhu di atas 40°C), sedangkan kubikel 3 diberi kondisi overhumidity (kelembaban > 70%). Hasil pengujian menunjukkan sistem mampu melakukan kontrol suhu dan kelembaban dalam ruang multi-kubikel serta mengirimkan notifikasi melalui Telegram dengan tingkat keberhasilan 100% dan rata-rata delay 5,6 detik.

Hijaiyah letters have varying shapes, and some of them are very similar, often causing errors in the manual character recognition process. This study aims to classify Hijaiyah letters based on digital images using the Convolutional Neural Network (CNN) method. This method was used in this study with a dataset consisting of 28 letter classes and a total of 4,480 images obtained from various public sources and private data. All images underwent a preprocessing stage that included labeling, resizing, normalization, and augmentation, then were divided into three parts, namely training data, validation data, and test data with a ratio of 70:20:10. The training process was carried out using the Python programming language with the help of the TensorFlow and Keras libraries on the Google Colab platform. The test results showed that the CNN model achieved an accuracy of 97.10%, with an average precision, recall, and F1-score of 0.97, respectively. Classification errors only occurred in letters that had similar shapes, such as Syin and Sin. Based on these results, the CNN method proved to be effective, efficient, and accurate in recognizing Hijaiyah letter image patterns, so it can be used as a basis for developing classification models with higher accuracy in the future.

Relational databases store information in interconnected tables and are widely used for data management and retrieval. However, in certain environments, the original values stored in a relational database cannot be exposed during data retrieval. This limitation creates a challenge because common encryption methods only transform data for storage and do not support mathematical operations needed for value matching. Partially Homomorphic Encryption is a cryptographic approach that allows specific mathematical operations to be performed directly on transformed data without restoring it to its original form. This study proposes the use of Partially Homomorphic Encryption to enable value-based data retrieval while keeping all stored values in their transformed form throughout the entire process. The method relies on homomorphic properties that allow mathematical comparison to be conducted on encrypted data, making the retrieval process possible without revealing the original values. The results show that this approach can perform data retrieval operations in a relational database while preserving the transformed structure of the stored data. The proposed method offers an alternative for environments that require data retrieval without exposing original values and demonstrates the potential of homomorphic techniques in supporting secure and functional data processing in relational database contexts.

Waste has become a serious environmental problem in Indonesia, which continues to increase along with population growth. The issue of waste management poses serious challenges for the environment, especially in the process of separating organic and inorganic waste. In the field of computer vision, recognising the type and shape of waste through camera images remains a challenge due to variations in shape, colour, and complex lighting conditions. Therefore, this problem utilises Deep Learning technology, which is expected to be widely applied in Indonesia, especially in large cities with high waste volumes. This study aims to distinguish between organic and inorganic waste using the Convolutional Neural Network (CNN) method based on digital images. The developed CNN model was trained to recognise the visual patterns of each type of waste and tested to measure its accuracy. The test results show that the CNN-based classification system is capable of achieving an accuracy rate of 95%, thus proving the effectiveness of this method in supporting artificial intelligence-based automatic waste sorting systems.

Indonesian Sign Language (BISINDO) serves as a primary communication medium for the deaf community; however, limited public understanding often creates barriers during daily interactions. This study aims to develop a real-time BISINDO word-level translation system using hand landmark extraction and temporal modeling with Long Short-Term Memory (LSTM). The system employs MediaPipe Hands to detect 21 hand landmarks per frame, which are then processed as sequential motion patterns to classify five BISINDO words: saya, terima kasih, maaf, nama, and kamu. A total of 250 gesture samples were recorded under controlled lighting conditions as the primary dataset. The processed sequences were used to train the LSTM model, which was subsequently integrated with an ESP32 microcontroller and a DFPlayer Mini module to produce direct audio output. Experimental results show that the model achieved an average accuracy of 86%, with precision and recall values ranging from 0.81 to 0.94. The confusion matrix analysis indicates that most gestures were correctly classified, although some errors occurred in gestures with similar initial motion trajectories. Integration testing demonstrated an average system latency of 3.8 seconds and an audio output success rate of 85%. These findings indicate that the proposed system is capable of translating BISINDO word-level gestures accurately, responsively, and consistently in real-time conditions. This study provides a strong foundation for the broader development of sign language translation systems, with potential enhancements in vocabulary expansion, multi-user datasets, and hardware optimization for deployment in real-world environments.