Reconstruction of reliable plasmaspheric electron density profiles is important for understanding physical processes in the plasmasphere. This paper proposes a technique that can be applied to correct the plasmaspheric electron density profiles using ray tracing by scrutinizing dispersion analyses of lightning whistlers. The Global Core Plasma Model and the International Reference Ionosphere were introduced as a reference electron density profile. Modifications of this electron density profile were then proposed to satisfy the dispersion characteristics of lightning whistlers measured by satellites in the plasmasphere. We first introduce two kinds of functions to modify the electron density: constant and linear, the linear function is more adequate. We applied our method to two lightning whistler events on 14 August 2017, measured by the Plasma Wave Experiment/Waveform Capture aboard the Arase satellite, and analyzed the dispersion of the observed lightning whistlers. We show how the density modification affects the delay time of the ray path and satisfies the dispersion characteristics under the appropriate adjustments.
The plasmasphere within Earth’s magnetosphere plays a crucial role in space physics, with its electron density distribution being pivotal and strongly influenced by solar activity. Very Low Frequency (VLF) waves, including whistlers, provide valuable insights into this distribution, making the study of their propagation through the plasmasphere essential for predicting space weather impacts on various technologies. In this study, we evaluate the performance of different deep learning model sizes for lightning whistler detection using the YOLO (You Only Look Once) architecture. To achieve this, we transformed the entirety of raw data from the Arase (ERG) Satellite for August 2017 into 2736 images, which were then used to train the models. Our approach involves exposing the models to spectrogram diagrams—visual representations of the frequency content of signals—derived from the Arase Satellite’s WFC (WaveForm Capture) subsystem, with a focus on analyzing whistler-mode plasma waves. We experimented with various model sizes, adjusting epochs, and conducted performance analysis using a partial set of labeled data. The testing phase confirmed the effectiveness of the models, with YOLOv5n emerging as the optimal choice due to its compact size (3.7 MB) and impressive detection speed, making it suitable for resource-constrained applications. Despite challenges such as image quality and the detection of smaller whistlers, YOLOv5n demonstrated commendable accuracy in identifying scenarios with simple shapes, thereby contributing to a deeper understanding of whistlers’ impact on Earth’s magnetosphere and fulfilling the core objectives of this study.
3 August 2025 in JOURNAL OF APPLIED INFORMATICS AND COMPUTING
Searching for relevant scientific literature faces complex challenges due to the proliferation of academic publications. This research develops a semantic similarity-based scientific paper recommendation system by utilizing Sentence Transformer (all-MiniLM-L6-v2 model) and cosine similarity algorithm on arXiv dataset (15,504 papers in Computer Science). The system is implemented as a Streamlit-based interactive web application that accepts user queries and recommends related papers based on semantic similarity. Performance evaluation using Precision, Mean Average Precision (MAP), Mean Reciprocal Rank (MRR), and Normalized Discounted Cumulative Gain (NDCG) metrics showed that embedding text from the Introduction section without pre-processing yielded the best performance (NDCG: 0.7590; MAP: 0.6960; MRR: 0.7254), outperforming Abstract-based or text combination approaches. A user test of 45 respondents confirmed the effectiveness of the system: 95.5% expressed satisfaction with the relevance of the recommendations, and 93.3% confirmed a significant reduction in manual search time. The findings prove that retaining the raw text structure in the Introduction is optimal for semantic representation. Development suggestions include multidomain dataset expansion and transformer model optimization for accuracy improvement.
31 December 2024 in SINTECH (Science and Information Technology) Journal
This study aims to develop a Vision Transformers (ViT) model for recognizing Korean characters (Hangeul) in response to the growing interest in learning the Korean language and Korean culture in Indonesia. The research methodology involves training the ViT model using a comprehensive dataset of 29,636 base Korean characters. The ViT model has achieved a significant level of accuracy with the score of 93% in recognizing base Korean characters. By integrating deep learning, this study is expected to make a positive contribution to the development of language learning tools for Korean character recognition, unlocking its potential for applications and systems based on the Korean language.
4 October 2025 in JOURNAL OF APPLIED INFORMATICS AND COMPUTING
High reading interest plays an important role in increasing knowledge and fostering a stronger literacy culture. With the growing access to information and technology, reading interest is also expected to improve through innovative and interactive platforms. However, traditional reading materials often fail to attract younger generations who are more engaged with digital content. To address this challenge, one of the efforts undertaken is the development of a modern platform that provides a collection of short stories enriched with cultural and educational values, tailored to appeal to contemporary readers. This study aims to design and implement a short story generation system using a Transformer-based language model, specifically T5 (Text-to-Text Transfer Transformer). The model is fine-tuned using a curated dataset of folktales from various regions, with the goal of producing relevant, engaging, and coherent narrative texts. The generation process is supported by pre-processing techniques to structure the data into narrative components such as introduction, conflict, climax, and resolution. The generated stories are then evaluated through human evaluation methods, including questionnaires and User Acceptance Testing (UAT), to assess their quality, coherence, engagement, and cultural relevance. This ensures that the system not only produces technically valid texts but also delivers narratives that are meaningful and enjoyable for readers. Ultimately, this study contributes to the promotion of literacy by presenting local wisdom and traditional values from diverse cultures through stories in a more modern, engaging, and accessible format for the younger generation.
19 April 2025 in Jurnal Ilmiah Merpati (Menara Penelitian Akademika Teknologi Informasi)
Indonesian, as the official language, follows the standards set by the Kamus Besar Bahasa Indonesia. Despite this, academic writing often contains errors such as non?standard word usage, typographical mistakes, and contextually inappropriate foreign terms. This study presents a web?based application for detecting and correcting non?standard words in scholarly documents, employing a Dictionary Lookup approach (referencing both the KBBI and English IT terminology) alongside the Damerau?Levenshtein Distance algorithm. Evaluation on 100 test entries yielded 100% detection accuracy, 98% precision, and 100% recall. Processing times averaged 0.03 seconds for 50 standard words and 57.13 seconds for 50 non?standard words. Incorporating frequency?based ranking improved the relevance of suggestions, with correct corrections most often appearing first. A maximum edit distance of two was identified as optimal, balancing accuracy and efficiency in User Acceptance Testing.