Sony Hartono Wijaya

Background: The rapid growth of online culinary platforms has produced a vast amount of unstructured Indonesian recipe texts. Extracting structured information such as ingredients, quantities, and measurement units002C is essential for downstream applications, including nutritional computation and diabetes-oriented dietary recommendation systems. However, Indonesian recipe texts present substantial challenges due to linguistic variability, informal writing styles, and limited availability of annotated corpora. Objective: This study investigates effective modeling strategies for Indonesian recipe Named Entity Recognition (NER) in low-resource settings and to identify reliable approaches to improve the robustness and generalization of extraction. Methods: A quantitative experimental approach was employed to evaluate the IndoBERT, BiLSTM-CRF, IndoBERT-CRF, and hybrid IndoBERT-BiLSTM-CRF architectures. A reliability-weighted voting ensemble combining heterogeneous models was proposed, and high-confidence ensemble predictions were used for pseudo-labeling in a semi-supervised setting. Performance was assessed using entity-level precision, recall, and F1-score. Results: IndoBERT achieved a strong baseline performance (F1-score = 96.61%). IndoBERT-CRF and the hybrid IndoBERT-BiLSTM-CRF model achieved competitive F1-scores (92.99% and 92.82%, respectively) but did not outperform the Transformer-only baseline. The proposed weighted ensemble achieved the best supervised performance (F1-score = 96.79%). Semi-supervised training with ensemble-generated pseudo-labels maintained competitive performance (F1-score = 96.72%) while improving recall and generalization to unseen ingredient expressions. Conclusion: Increased architectural complexity does not necessarily improve performance on the Indonesian recipe NER task. Reliability-weighted ensemble learning combined with pseudo-labeling provides a stable and effective approach for low-resource culinary NER and supports future nutrition-oriented recipe processing. Keywords: Indonesian Recipe Text; Pseudo-Labeling; Named Entity Recognition; Semi-Supervised Learning; Weighted Ensemble

Near-infrared (NIR) spectroscopy is a powerful analytical technique, but its effectiveness is often hindered by instrumental noise and a dependency on expert-driven preprocessing and calibration. This study introduces a novel, intelligent deep learning architecture, a deep multi-block convolutional neural network (CNN) that addresses these challenges by performing end-to-end quantitative analysis directly on raw spectral data. By segmenting the NIR spectrum into multiple blocks and processing them in parallel, the model mimics multi-scale feature learning, automatically identifying and leveraging localized spectral regions relevant to target properties. This approach eliminates the need for manual preprocessing, a significant step toward creating autonomous analytical systems. This study evaluated the model on four rice quality datasets (moisture, protein, fatty acids, and total fat) and benchmarked it against traditional partial least squares regression (PLSR) and other deep learning models. The proposed multi-block CNN consistently achieved superior performance, demonstrating its ability to handle raw spectral data robustly. For instance, it achieved coefficient of determination (R2) values of 0.996 ±0.003 for total fat and 0.997 ± 0.001 for fatty acids. Statistical tests confirm that the model’s performance on raw data is not significantly different from its performance with optimized preprocessing, validating its potential for real-world artificial intelligence applications in agriculture, food quality monitoring, and biosystems engineering where rapid, automated analysis is critical.

Buffaloes are important in animal husbandry, agriculture, and sociocultural and religious activities in Indonesia. The buffalo population has decreased at the national and regional levels, including in the Batanghari District, Jambi Province, Indonesia. This study analyzed the population structure, effective population size, and inbreeding rate of buffalo populations in the Batanghari District, Jambi, Indonesia, based on secondary data. The data population of 3,149 buffaloes used in this study was sourced from the Integrated National Animal Health System (ISIKHNAS) in the Batanghari District in 2023. The results showed a calf crop of 21.71%, a calving rate of 16.61%, a natural increase of 14.74%, and a net replacement rate of 279.51%. The effective population size was 592 heads, and the inbreeding rate was 0.08%. It can be concluded that the natural increase rate of the buffalo population in the Batanghari District was low, but the number of young replacement animals was sufficient. The effective population size was 592 heads, and the level of inbreeding per generation remained within acceptable limits. Although the buffalo population in the Batanghari District exhibited a negative trend, it still had potential as a source of breeding stock, as indicated by the replacement rate.

Short Message Service (SMS) is a widely used communication media. Unfortunately, the increasing usage of SMS has resulted in the emergence of SMS spam, which often disturbs the comfort of cellphone users. Developing a classification model as a solution for filtering SMS spam is very important to minimize disruption and loss to cellphone users due to SMS spam. To address this issue, utilize the Naïve Bayes algorithm and Support Vector Machine (SVM) along with Chi-square and Information Gain. This study focuses on the classification and analysis of SMS spam on a cellular operator service in a telecommunications company using machine learning techniques. This study applies and combines a combination of classification methods including Naive Bayes and Support Vector Machine (SVM). The combination is carried out with Chi-square and Information Gain feature selection to reduce irrelevant features. This study also applies a combination with data balancing techniques using the Synthetic Minority Oversampling Technique (SMOTE) to balance the number of unbalanced classes. The results show that SMOTE improves classification performance. SVM performs spam SMS classification or not spam Model 7 (SVM) achieves accuracy 98,55% and it has improved the performance when it was combined with SMOTE Model 10 (SVM + SMOTE) achieves F1-score 99,23% in performing spam SMS classification or not this outperforms all other models. These results indicate that the SVM algorithm achieved better performance in detecting spam SMS compared to Naive Bayes, which demonstrated a lower level of accuracy. These results illustrate the effectiveness of combining machine learning models to enhance classification accuracy with balanced data, emphasizing the model that exhibited the most substantial improvement in performance.

Honey bees have various types and characteristics, one of which is the stingless bee. This type has limitations in producing honey, the selling value of its nest is quite expensive, and the water content in the honey produced is relatively high. The high water content affects the shelf life of this type of honey product, making it a challenge for honey farmers in marketing it. In addition, the dominant sour taste also makes its market increasingly limited or vulnerable to falsification of its purity by irresponsible producers. The use of spectrophotometers is increasingly developing in the food sector, especially in detecting the purity of a food product. The portable type of spectrophotometer also makes it easier to obtain spectrum data for a particular product. A simpler technique that is directly connected to a computer device allows it to be developed into a cloud-based application by providing minimal raw data processing (pre-processing). This study produces an android-based application and a simple cloud-based application architecture, which aims to facilitate the application of a honey purity prediction model from stingless bees. The Android-based application was successfully created by applying 'raw' spectrum data processing from the results of scanning a portable spectrophotometer, and data experiments with the SVM classification model produced an accuracy of 95%. The application of PCA techniques to cloud-based mobile application architecture results in efficient preprocessing of spectrum data.

The Complex underwater characteristics, challenges for image processing tasks. These images often have poor visibility due to low contrast, light scattering and various types of interference. There is a lack of exploration into the effectiveness of existing underwater image enhancement methods, particularly in the context of seagrass ecosystems, allows for further investigation. This study aims to explore and evaluate the effectiveness of various methods in underwater image enhancement, including Colour Balanced, CLAHE, and Unsharp Masking and their combinations, starting with converting video data from UTS devices into two-dimensional images. Furthermore, the quality of images taken from underwater cameras placed in a complex and wild seagrass meadow environment was improved using the proposed method, and the quality was evaluated by the SSIM value. The results show that the CLAHE method has the highest average SSIM value of 0.898. Meanwhile, the combined Color Balanced-CLAHE method achieved an SSIM value of 0.683 in a separate evaluation. This combination is an innovative approach to address complex underwater image quality problems, providing a more specific and adaptive solution. Overall, the proposed method is able to improve the visual quality of images on aspects such as clarity, color, and visibility of objects in the image

The Unani Tibb is a medical system of Greek descent that has undergone substantial dissemination since the 11th century and is currently prevalent in modern South and Central Asia, particularly in primary health care. The ingredients of Unani herbal medicines are primarily derived from plants. Our research aimed to address the pressing issues of antibiotic resistance, multi-drug resistance, and the emergence of superbugs by examining the molecular-level effects of Unani ingredients as potential new natural antibiotic candidates. We utilized a machine learning approach to tackle these challenges, employing decision trees, kernels, neural networks, and probability-based methods. We used 12 machine learning algorithms and several techniques for preprocessing data, such as Synthetic Minority Over-sampling Technique (SMOTE), Feature Selection, and Principal Component Analysis (PCA). To ensure that our model was optimal, we conducted grid-search tuning to tune all the hyperparameters of the machine learning models. The application of Multi-Layer Perceptron (MLP) with SMOTE pre-processing techniques resulted in an impressive accuracy precision and recall values. This analysis identified 20 important metabolites as essential components of the formula, which we predicted as natural antibiotics. In the final stage of our investigation, we verified our prediction by conducting a literature search for journal validation or by analyzing the structural similarity with known antibiotics using asymmetric similarity.

The demand for halal products in the Islamic context continues to be high, requiring adherence to halal and haram laws in consuming food and beverages. However, individuals face the challenge of distinguishing between haram meat and permissible halal meat. This study aims to answer these challenges by designing an expert system application within the ERP framework to increase the usability functionality of the system that can differentiate between beef, pork, or a mixture of both based on the physical characteristics of the meat. The aim is to determine halal products permissible for consumption by Muslims. The research methodology includes a data collection process that involves taking 30 meat samples from various sources, and the criteria used to classify the meat will be determined based on an analysis of the physical characteristics of the meat. System administrators use expert systems to ensure proper treatment of meat during administration processes, including separating halal beef from pork and implementing different inventory procedures. The Certainty Factor (CF) inference engine deals with uncertainty even though the expert system's accuracy level is relatively good with several rules. However, these results must be studied further because the plan relies on expert opinion. Therefore, it is necessary to set the correct CF value for accurate height classification. The CF inference engine facilitates reasoned conclusions in meat classification. Functional testing confirms the smooth running of the system, validating its reliability and performance. In addition, the expert system accuracy assessment produces a commendable accuracy rate of 90%. In addition, the expert system works powerfully on various meat samples, accurately classifying meat types with high precision. This study explicitly highlights the expert system's design for meat classification in determining halal products by using the Expert System Certainty Factor. In conclusion, this expert system provides an efficient and reliable approach to classifying meat and supports the production and consumption of Halal products according to Islamic principles.

Researchers and journal managers need summary information, such as research maps and trends. Topic and words-based document content analysis alternative to science mapping and trend prediction based on bibliographic analysis. The data are a collection of journal articles/proceeding documents and metadata for 2011-2020 published by the International Journal of Electrical and Computer Engineering (IJECE). A combination of several techniques and methods is used, such as text mining, topic modeling, cosine similarity, network analysis, graph theory, and seasonal autoregressive integrated moving average (SARIMA). This research has produced research topics, mapping, trend prediction, and visualization according to its objectives. The results of the topic coherence test obtained the optimal number of topics, as many as 6. The results of the topics were evaluated by experts to be given labels and areas of focus. On the research map for each topic, information is found on trending research, the most popular research, research that is central to the research group, and critical research on the development of the group path. It also identifies the type of breakthrough, incremental, and research gap. Predictions of research trends obtained are based on topics and words that describe the development of research. Visualization is descriptive and predictive.