eISSN : 3022-7011
ISSUER : KIPS
 
After the Korea Information Processing Society (KIPS) Transactions journal was founded in 1994, it was reorganized into the KIPS Transactions: Computer and Communication Systems(2287-5891/2734-049X ) and the KIPS Transactions: Software and Data Engi neering(2287-5905/2734-0503) in 2012. Through the KIPS official meeting on January 8th, 2024, the new KIPS Transaction journal was founded by integrating two KIPS Journals, KIPS Transactions: Computer and Communication Systems and KIPS Transactions: Software and Data Engineering. The new journal aims to realize social value and contribute to the development of South Korea’s science and technology with support from the lottery fund of the Ministry of Strategy and Finance and the science/technology promotion fund of the Ministry of Science and ICT. It is indexed in the Korea Science Academic Database, Korea Citation Index (KCI), and EBSCO.

HighlightsMore

Diffusion-based Audio-to-Visual Generation for High-Quality Bird Images

Adel Toleubekova  Joo Yong Shim  XinYu Piao  Jong-Kook Kim
Accurately identifying bird species from their vocalizations and generating corresponding bird images is still a challenging task due to limited training data and environmental noise in audio data. To address this limitation, this paper introduces a...

Bambda: A Framework for Preventing Function Invocation Condition-Based Attacks in Serverless Environments

Shin Chang Hee  Lee Seung Soo
Serverless computing is rapidly emerging as a new paradigm in cloud computing, offering automatic scalability, cost efficiency, and ease of operation. However, its two core characteristics—IAM-based privilege management and event-driven execution—ca...

PEFT Methods for Domain Adaptation

Lee You Jin  Yoon Kyung Koo  Chung Woo Dam
This study analyzed that the biggest obstacle in deploying Large Language Models (LLMs) in industrial settings is incorporating domain specificity into the models. To mitigate this issue, the study compared model performance when domain knowledge wa...

Label Differential Privacy Study for Privacy Protection in Multimodal Contrastive Learning Model

Youngseo Kim  Minseo Yu  Younghan Lee  Ho Bae
Recent advancements in multimodal deep learning have garnered significant attention from both academia and industry due to their exceptional accuracy and ability to learn rich knowledge representations. In particular, contrastive learning based appr...

Latest Publication   (Vol. 15, No. 1, Jan.  2026)

AI-Based Automated Framework for Online Apparel Product Detail Pages Generation
Joo Yong Shim  JiYoon Park  Yerim Choi
Product Detail Pages (PDPs) involves significant manual efforts, resulting in high costs, time consumption, dependence on skilled labor, and various technical limitations. To address these challenges, this study proposes an AI-based integrated automation framework for PDPs generation. The proposed framework standardizes PDPs into four areas and applies automation algorithms, particularly to the ‘product feature appeal’ and ‘optional photo arrangement’ sections, thereby enhancing overall page generation efficiency. In the ‘product feature appeal’ section, a large language model pre-trained on fashion domain data is used to automatically generate text. This framework also adopt model that is trained to automatically select images that matches the generated text. In the ‘photo arrangement by product options’ section, images are segmented into foreground and background, and embeddings are generated. Hierarchical clustering, reflecting both foreground and background information, is then used to group similar images. By analyzing frame types, images in the clusters are arranged to display optional products and model images naturally. The proposed framework is validated using real-world data. The results demonstrate that the proposed method excels in text generation, image clustering, and image arrangement consistency. Furthermore, evaluations by online shopping mall operators, related experts, and consumers shows that quality level of generated PDP is comparable to manual work. This study presents the first automation framework that comprehensively addresses both text and image aspects, thereby proving its applicability and scalability in the fashion industry.
The Transactions of the Korea Information Processing Society, Vol. 15, No. 1, pp. 1-11, Jan. 2026
10.3745/TKIPS.2026.15.1.1
AI-Based Automation Automated Framework Product Detail Pages Generation clustering Text-Generation
A Finite Memory Structure Smoothing Filter Based Road Vehicle’s Suspension System with Temporary Uncertainties
Pyung Soo Kim
In this paper, an alternative road vehicle's suspension system with temporary uncertainties is developed. The finite memory structure(FMS) smoothing filter is adopted to provide intrinsic robustness as well as to consider cases where there is a fixed delay between a measurement and the availability of its estimate. The single-corner model, which is also called the 1/4 model or the quarter-car model for one of the four wheels, for the road vehicle's suspension system and its state-space model are described. Model uncertainty and unknown input are introduced as representative temporary uncertainties, and the role of estimation filter is described to resolve problems due to temporary uncertainties. Then, the FMS smoothing filter based road vehicle's active suspension system is developed under temporary uncertainties. The time-invariance of the FMS smoothing filter gain is explained and the relationship between the window length and the performance is briefly described. To verify the performance of the developed FMS smoothing filter based system, extensive computer simulations using MATLAB are performed to cover both nominal and temporarily uncertain systems. It is shown that the FMS smoothing filter based road vehicle's active suspension system can outperform not only FMS filter based system but also infinite memory structure(IMS) filter based system. Moreover, after the effect of temporary uncertainty completely disappear, the FMS smoothing filter based system is shown to be comparable to FMS filter based and IMS smoothing filter based systems.
The Transactions of the Korea Information Processing Society, Vol. 15, No. 1, pp. 12-20, Jan. 2026
10.3745/TKIPS.2026.15.1.12
Road Vehicle Suspension System FMS Filter IMS Filter Intrinsic Robustness Temporary Uncertain System
Predictive Data Analysis-Based Modeling of Electric Vehicle Future Driving Range
Heeseo Jeong  Dajeong Lee  Kyujin Cho  Charmgil Hong
As interest in eco-friendly transportation methods increases, battery degradation in Electric Vehicles (EVs) poses a critical barrier to the adoption of EVs. Our research aims to predict the range of EVs one year into the future and use this prediction as an indicator to diagnose the degree of battery degradation. Battery Management System data collected from 371 vehicles over two years underwent preprocessing and feature engineering before being used for Feature Tokenizer Transformer model training. The Feature Tokenizer Transformer has structural advantages that enabling effective capture of complex relationships among all input variables regardless of variable types, which makes it well-suited for processing tabular data such as Battery Management System data. In actual EV operation, most cases involve charging before the battery is completely discharged, which makes it difficult to collect ground truth values of range required for model training. To overcome this constraint, we propose a method that predicts energy efficiency one year ahead and subsequently converts it to range. Through a stepwise prediction framework divided into energy efficiency prediction and range calculation, we can predict range without complete discharge data from EV batteries. This research achieved a prediction accuracy of Mean Absolute Error 8.872(km), which confirms that our methodology can be effectively applied to future range prediction problems based on real-world collected data.
The Transactions of the Korea Information Processing Society, Vol. 15, No. 1, pp. 21-27, Jan. 2026
10.3745/TKIPS.2026.15.1.21
Electric vehicle Battery Management System Range Prediction Battery Degradation Feature Tokenizer Transformer
Integrated Analysis for Knowledge Graph Embedding and Text Embedding
Chun-Hee Lee
Knowledge graphs have been utilized in a broad range of areas such as recommendation systems and question-answering systems. A knowledge graph can contain not only graph structures that represent the connections between entities but also the detailed explanations for each entity. Graph structures and explanations can be easily applied to various areas using knowledge graph embedding and text embedding, respectively. Although a knowledge graph can contain multi-modal data such as graph structures and texts, there has been little research on the integrated analysis of embeddings generated from each modality. This paper compares and analyzes knowledge graph embedding and text embedding on the YAGO3-10 data. Principal component analysis and correlation analysis are first performed on knowledge graph embedding and text embedding data. Then, the relationships with degrees, which are one of the important characteristics in a graph, are analyzed. Finally, this paper constructs linear regression models between knowledge graph embedding and text embedding.
The Transactions of the Korea Information Processing Society, Vol. 15, No. 1, pp. 28-36, Jan. 2026
10.3745/TKIPS.2026.15.1.28
Knowledge Graph Embedding text embedding integrated analysis multi-modality
Quantitative Model Design for Optimizing the Difficulty of Cognitive Games
Kim Se Yeon  Lim Young Hoon
This study proposes a novel Difficulty Index (ID) model to quantitatively assess cognitive load using cognition-based games and validates it through comparison with existing alternative models. The experiment required participants to perform a task in which they entered digits from 0 to 9 in reverse order according to the presentation speed. A total of 60 difficulty conditions were designed by combining the number of digits and display time. The proposed model was defined to reflect the limitations of working memory capacity, mathematically expressing the assumption that difficulty increases exponentially when the number of digits exceeds four. For comparison, a simple linear model and a logarithmic-scale model adapted from Fitts’ Law were employed. Regression analysis revealed that the proposed model exhibited the highest explanatory power and demonstrated superior predictive performance compared to the other two models. These findings suggest that the proposed model effectively captures the nonlinear increase in task difficulty and can serve as a reliable indicator for predicting actual user performance. Furthermore, by integrating theoretical foundations of working memory with experimental validation, this study provides a practical tool for designing educational and training environments that balance challenge and learning efficiency, thereby offering both academic and practical contributions.
The Transactions of the Korea Information Processing Society, Vol. 15, No. 1, pp. 37-45, Jan. 2026
10.3745/TKIPS.2026.15.1.37
Difficulty Index working memory Regression Analysis Cognitive Load Cognitive Games
A Study on the Classification Performance of Machine Learning-Based Zentner Music Emotion Model
Chung Moonsik  Moon Nammee
This study explores music emotion classification using machine learning. The emotional categories are based on Zentner's music emotion model, and labels from this model were assigned to audio data from each genre in the GTZAN dataset. To perform classification, audio features such as MFCC, ZCR, Chroma, Spectral Centroid, and Harmony were extracted, and machine learning models including KNN, Decision Trees, Random Forest, XGBoost, LightGBM, and SVM were used to compare classification performance. In addition, the effectiveness of Recursive Feature Elimination and stacking techniques was tested, and improvements in classification performance were observed through stacking. The experimental results showed that among individual models, LightGBM achieved the highest classification performance. Moreover, applying the stacking technique led to a 14% improvement in performance on average compared to individual models.
The Transactions of the Korea Information Processing Society, Vol. 15, No. 1, pp. 46-53, Jan. 2026
10.3745/TKIPS.2026.15.1.46
Music Emotion Zentner Model Machine Learning Recursive Feature Elimination stacking
NTP-Based DPO Training for Suppressing Korean-Chinese Code-Switching
Sungmin Ko  Youhyun Shin
Large Language Models (LLMs) have demonstrated outstanding performance across various natural language processing (NLP) tasks. However, when Chinese-pretrained LLMs are applied in Korean environments, unintended code-switching between Korean and Chinese frequently occurs. This not only undermines user trust but also poses critical issues in domains requiring linguistic accuracy, such as translation, education, and official document writing.To address this problem, we propose a Direct Preference Optimization (DPO) training method based on Next Token Prediction (NTP). Specifically, we leverage NTP to detect confusion points—the moments when Chinese tokens first appear in Korean outputs. Using these points, we construct the NTP-CS dataset by pairing responses with code-switching (rejected) against responses without code-switching (chosen), and train the model accordingly. Experimental results show that our proposed NTP-CS approach consistently outperforms both LLM-CS, which induces code-switching via LLM prompts, and LLM-TX, which translates entire sentences. Notably, across all datasets, chosen responses exhibited positive log-likelihood values, while rejected responses showed negative values, forming an ideal probability pattern. This demonstrates that datasets constructed around local confusion points are more effective than simple translation datasets for mitigating code-switching. We expect that this research will improve the consistency of Korean-centric outputs in multilingual settings and help reduce the quality gap between Korean and Chinese responses.
The Transactions of the Korea Information Processing Society, Vol. 15, No. 1, pp. 54-60, Jan. 2026
10.3745/TKIPS.2026.15.1.54
Artificial intelligence Natural Language Processing Code Switching large language models
Enhancing LLM-driven Program Repair through RAG and Test-Method Mapping
Aslan Safarovich Abdinabiev  Eunseo Jung  Byungjeong Lee
Automated Program Repair (APR) aims to automatically fix buggy programs, with recent advances in Large Language Models (LLMs) showing strong potential. However, existing methods often struggle to leverage contextual information and capture the relationships between buggy methods and failing test cases. This paper aims to enhance LLM-driven program repair by systematically integrating intelligent method-test mapping with retrieval-augmented generation to address these limitations. We present an LLM-driven framework with three key components: first, a mapping strategy that links buggy methods to relevant test cases, next, retrieval-augmented generation (RAG) that incorporates semantically similar code examples, and lastly, an iterative feedback-driven generation process. We constructed the framework using GPT-4o and evaluated it on 353 method-level bugs from Defects4J 1.2. Experimental results demonstrated that our approach efficiently repaired 181 bugs, outperforming the baseline methods through effective context utilization and guided patch generation.These results indicate that retrieval-enhanced contextual reasoning can substantially improve repair accuracy and scalability.
The Transactions of the Korea Information Processing Society, Vol. 15, No. 1, pp. 61-69, Jan. 2026
10.3745/TKIPS.2026.15.1.61
Automated Program Repair large language models Retrieval-Augmented Generation
Physics-Informed Estimation of Nacelle Free-Stream Wind SpeedUsing SCADA Measurements
Taehyoung Kim  Jiman Hong
The performance evaluation of wind turbines strongly depends on the accurate estimation of free-stream wind speed. However, the nacelle wind speed commonly used in practice is typically lower than the actual atmospheric wind speed due to rotor induction and wake effects, which reduces the accuracy of aerodynamic analysis and digital twin models. Therefore, this paper proposes a physics-based correction method that estimates the free-stream wind speed from nacelle measurements by combining wind turbine SCADA data with the drivetrain power balance equation. The proposed method formulates the power balance equation as a nonlinear root-finding problem and applies the Brentq algorithm along with heuristic rules to correct the bias caused by rotor induction and wake effects. Evaluation results for both low- and high-wind-speed regions show that the estimated wind speed achieves improved accuracy across all metrics - RMSE, MAE, NSE, and CCC - compared to the nacelle wind speed, and maintains stable performance even in high-wind-speed regions where pitch control is active.
The Transactions of the Korea Information Processing Society, Vol. 15, No. 1, pp. 70-74, Jan. 2026
10.3745/TKIPS.2026.15.1.70
wind turbine Wind speed estimation Drivetrain Power Balance Physics-based method Root-finding algorithm

Forms

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)

Old Journals

Indexing

All publications of TKIPS is indexed in DOI, EBSCO, Google Scholar, Crossref, and CrossCheck.

TKIPS is also selected as the Journal for Accreditation by NRF (National Research Foundation of Korea).

Related Journals