Digital Library

Vol. 21, No. 6, Dec. 2025  Cover

1. Clustered Federated Learning Based on Mahalanobis Distance for Sequential Medical Data
   Tae Hwan Yoon, Bong Jun Choi
   Vol. 21, No. 6, pp. 564-574, Dec. 2025  
   https://doi.org/10.3745/JIPS.03.0211
   Details Full Text PubReader
   Keywords: clustering, Detecting Emotion and Stress, Federated learning, Mahalanobis distance
   Show / Hide Abstract

   In hospitals, metadata typically contains patients' personal information based on the doctor's diagnosis. Therefore, sniffers or hijackers could launch attacks to steal important information from hospitals or patients. For this reason, hospital data must be anonymized and protected by specialized systems to ensure its safe use, especially when multiple hospitals share data. If hospitals implement systems that can securely share data while maintaining privacy, researchers and clinicians can leverage large amounts of distributed data to more effectively train deep learning models. In this context, we select a solution based on clustered federated learning (CFL). In typical CFL scenarios, forming appropriate clusters can help build more personalized models for different groups. However, previous CFL approaches still face challenges from model heterogeneity. To further mitigate the heterogeneity problem, we propose a Mahalanobis distance based clustered federated learning (MD-CFL) method, which offers advantages in reducing model heterogeneity and improving clustering performance by correcting for feature skew in non-normalized data. Our experiments show that MD-CFL achieves accurate clustering performance, with a higher silhouette score compared to cosine-based FedAvg.


2. Research on Risk Assessment of Internal Audit Information in Enterprises through Data Mining
   Lu Xia, Tiantian Wang
   Vol. 21, No. 6, pp. 575-584, Dec. 2025  
   https://doi.org/10.3745/JIPS.04.0361
   Details Full Text PubReader
   Keywords: Audit Information, Data Mining, Deep Learning, Risk Identification
   Show / Hide Abstract

   This paper initially segmented the audit information samples using a clustering algorithm. Subsequently, a backpropagation neural network (BPNN) algorithm enhanced by the beetle antennae search (BAS) algorithm was used for risk assessment. Financial report data was crawled from listed companies for a case analysis to assess the impact of the number of clustering centers on the clustering algorithm and the effect of the activation function type on the improved BPNN algorithm. Additionally, the audit information risk identification performance of the support vector machine (SVM), traditional BPNN, sparse autoencoder-BPNN, and the improved BPNN algorithm was compared. The findings revealed that the clustering algorithm demonstrated optimal sample division performance when utilizing two clustering centers. Moreover, the improved BPNN algorithm exhibited superior performance under the sigmoid activation function, outperforming both SVM and traditional BPNN algorithms.


3. Triple-Cascade Prompt-Based Implicit Sentiment Analysis with Data Augmentation and Automated Feedback Correction
   Yu Zhang, Xu Li, Shuxin Chen, Jiaqi Wang
   Vol. 21, No. 6, pp. 585-597, Dec. 2025  
   https://doi.org/10.3745/JIPS.02.0230
   Details Full Text PubReader
   Keywords: data augmentation, Prompt Learning, Multi-hop Reasoning, Self-Feedback Correction
   Show / Hide Abstract

   To address the challenges of implicit aspect-based sentiment analysis, specifically the lack of explicit sentiment expressions and the complexity of text semantics, we introduce a three-stage cascaded prompting reasoning model for implicit sentiment analysis based on data augmentation and automatic feedback correction. Initially, the model enhances the semantic information of the text by extracting target word concept representations from external knowledge bases, while also leveraging syntactic desensitization transformations to enhance syntactic information. Subsequently, we construct prompt templates and concatenate them with the enhanced text, inputting the result into a T5 model to infer target aspect words, implicit opinion expressions, and sentiment polarity in a stepwise manner. Finally, the model employs a large language model to self-correct the inference results, further improving the accuracy of the analysis. Experimental results demonstrate that the proposed model achieves F1-scores of 74.63% and 76.17% on the Restaurant and Laptop datasets, respectively. Compared to mainstream implicit aspect-based sentiment analysis models, this represents an improvement of 2.35% and 0.58%. These findings validate the effectiveness of the proposed model in the implicit aspect-based sentiment analysis task.


4. Design and Development of a Summarization Service Prototype of Personal Health Record Medical Data with Fast Healthcare Interoperability Resources Structure Using Large Language Models
   Nam-Gyu Lee, Seung-Hee Kim
   Vol. 21, No. 6, pp. 598-612, Dec. 2025  
   https://doi.org/10.3745/JIPS.04.0362
   Details Full Text PubReader
   Keywords: electronic medical record, Fast Healthcare Interoperability Resources, Generative Artificial Intelligence, Large Language Model, Personal health record, ChatGPT
   Show / Hide Abstract

   As the scope of healthcare data expands beyond hospital-generated electronic medical records (EMRs) to include personal health records (PHRs), there is a growing need for automated methods to efficiently summarize large volumes of patient-specific information. In this study, we propose a summarization approach that leverages large language models (LLMs) and standardized data formats to improve accessibility and usability of patient data. Specifically, we developed a prototype system that summarizes patient Bundles formatted in accordance with the Fast Healthcare Interoperability Resources (FHIR) standard. Using the ChatGPT API and document processing techniques, we generated summaries and evaluated their accuracy using a checklist based on clinical criteria. The summarization model achieved an accuracy of 81.6%, suggesting its potential for real-world application. Our findings indicate that healthcare professionals can more quickly and effectively review a patient’s primary conditions using summarized PHR data, particularly as FHIR adoption increases. However, the results also highlight certain limitations, including the generalization of summaries and the absence of domain-specific fine-tuning. These findings underscore the importance of future research involving multidisciplinary clinical evaluations, targeted fine-tuning strategies, and question-driven summarization to enhance accuracy and clinical relevance. Overall, this study demonstrates the feasibility of integrating LLM-based summarization into healthcare workflows, contributing to improved interoperability and decision-making in clinical settings.


5. Base View Rate Control Algorithm Based on VVC for 3D Video Sequences
   Tao Yan, Qian Zhang
   Vol. 21, No. 6, pp. 613-623, Dec. 2025  
   https://doi.org/10.3745/JIPS.02.0231
   Details Full Text PubReader
   Keywords: Scene Switching, Structure Similarity, VVC, 3D video coding
   Show / Hide Abstract

   A new generation of versatile video coding (VVC) standards was released in 2020; however, existing rate control algorithms for three-dimensional (3D) video coding based on VVC do not consider the effects of scene switching in actual video applications. Under the background of scene switching, the quality of encoded images degrades owing to the unreasonable allocations of coding resources. To solve this problem, this study proposes a 3D video basic view rate control algorithm based on scene switching detection of image structure similarity. First, scene switching is detected before coding. Basic view target bits estimated by the allocation model are first estimated by view weight factors. Finally, scene switching is judged by combining the structural similarity between frames with its transformation trend. When scene switching occurs, the rate control parameters and coding structure are adjusted in time. Experimental results show that the proposed algorithm not only tracks the target bit rate accurately but also significantly improves the reconstructed image quality in the case of scene switching and ensures that the decoder input buffer does not overflow or underflow.


6. Automated Proofreading of Chinese Text Using Syntactic Analysis and Semantic Layering
   Lili Zhu
   Vol. 21, No. 6, pp. 624-637, Dec. 2025  
   https://doi.org/10.3745/JIPS.04.0363
   Details Full Text PubReader
   Keywords: Automated Proofreading, Chinese Text, Semantic Layering, syntactic analysis
   Show / Hide Abstract

   This paper presents a study that integrates two primary methodologies for investigating automated proofreading: one employing a state-of-the-art syntactic analyzer and another based on a sophisticated semantic hierarchical model. The analyzer meticulously processed the input text, extracting part-of-speech tags and intricate syntactic dependencies, while the semantic hierarchical model was used to perform an innovative analysis. The resulting integration scheme of syntactic rigor and semantic depth represents a paradigm shift in error detection, outperforming the benchmarks established in the literature and state-of-the-art software systems. The integration of the syntactic structure with semantic understanding resulted in a marked increase in error detection accuracy. In particular, the study demonstrated a substantial enhancement in the average F-measure, surpassing Kingsoft WPS (2024) by 39% and Microsoft Word (2024) by 55%. It is worth noting that for error types that have historically been difficult to improve F-measure, particularly word ambiguity, the study achieved a 65% increase in detection accuracy.


7. Combined Query Based Image Retrieval Using Structure Elements Descriptor
   Gunho Lee, Minjoong Jeong
   Vol. 21, No. 6, pp. 638-650, Dec. 2025  
   https://doi.org/10.3745/JIPS.02.0232
   Details Full Text PubReader
   Keywords: Global Descriptor, Image retrieval, Structural Element Descriptor
   Show / Hide Abstract

   Similar image retrieval involves identifying and ranking images from a database based on visual attributes such as color, texture, and shape, with the goal of finding those most closely matching a given query image. This task requires precise analysis of image content to achieve accurate results. In this study, we propose an approach that incorporates structural information derived from an image segmentation model. This structural information highlights image characteristics, such as object shapes and their backgrounds, which are not fully captured by traditional dense global descriptors. By combining this structural information with global descriptors, our method captures both detailed shapes and broader image features in a user-controllable manner. Experimental results demonstrate the effectiveness of this integration approach in improving the performance of similarity search tasks.


8. Attention-Based Relational Learning Model for Few-Shot Knowledge Graph Completion
   Yuanxia Zhang, Hua Li, Yu Chen, Daoqing Gong
   Vol. 21, No. 6, pp. 651-663, Dec. 2025  
   https://doi.org/10.3745/JIPS.01.0114
   Details Full Text PubReader
   Keywords: Few-Shot Knowledge Graph Completion, knowledge representation, Meta-Learning, Relation-Centric Learning
   Show / Hide Abstract

   Knowledge graphs are crucial for numerous applications, but their frequent incompleteness limits their utility. Few-shot knowledge graph completion (FKGC) addresses this by learning to infer new facts from only a handful of examples. However, existing FKGC methods are highly vulnerable to noisy or inconsistent reference examples, which can severely degrade model performance. To overcome this critical challenge, we introduce ATMR, an attention-based meta-relational learning framework. ATMR incorporates a novel attention mechanism that strategically identifies and upweights the most informative reference triples while diminishing the influence of potential noise. This allows for the construction of more robust and accurate relation representations. Rigorous experiments on two public datasets demonstrate that ATMR consistently outperforms baselines. Notably, it achieves an 8.5% improvement in the Hits@10 metric for 5-shot completion on the NELL-One dataset, validating its superior ability to handle noise in few-shot scenarios.