Selected work

Representative Papers

A feasibility study on filtering low-accessibility web pages considering color vision deficiency

Jun 20, 2026

This study addresses the challenge of color accessibility for users with color vision deficiencies when browsing the web by proposing a machine learning–based automatic filtering approach. It pioneers the use of predictive modeling to identify and exclude webpages that violate Color Universal Design (CUD) principles, thereby exhibiting low accessibility. The method integrates CUD guidelines into a tailored evaluation metric, and experiments on 21 real-world webpages demonstrate that the model achieves a maximum AUC of 0.76, confirming the feasibility of automatically enhancing web color accessibility. This work offers a scalable technical pathway and practical insights for improving information accessibility for individuals with color vision deficiencies.

0 citationsRead paper

Exploring the Relationship Between Local Election Results and Online Public Opinion in Taiwan: A Case Study of Taitung County

Jan 06, 2026arXiv.org

This study examines local elections in Taitung County, Taiwan, to investigate the relationship between online public sentiment and actual voting outcomes, addressing a gap in empirical research on this topic. By systematically analyzing the correlation between social media discourse volume during the election period and candidates’ vote shares, the research establishes an empirical framework linking digital舆情 to voter behavior. Findings indicate that online discussion can, to a certain extent, serve as a reliable predictor of electoral results, offering new insights and practical implications for political communication, campaign strategy optimization, and election forecasting in the digital era. The study also acknowledges limitations concerning data representativeness and methodological constraints.

0 citationsRead paper

An Empirical Study on User Profile Analysis and SEO Performance: A Case of Taiwan Cultural Memory Bank 2.0

Jan 06, 2026arXiv.org

This study aims to enhance the digital outreach effectiveness and service quality of Taiwan’s Cultural Memory Bank 2.0 platform. By integrating the Visitor Relationship Management (VRM) framework, user behavior analytics, and Search Engine Optimization (SEO) assessment, the research empirically examines user demographics, browsing patterns, and engagement behaviors while quantifying the platform’s search visibility and organic traffic performance. Key user segments and their behavioral characteristics are identified, and combined with SEO metrics to formulate data-driven strategies for website optimization and social media promotion. The primary contribution lies in the synergistic integration of user personas with SEO indicators, offering actionable pathways to improve both user experience and online visibility for digital cultural platforms.

0 citationsRead paper

Deep Learning-Based Image Recognition for Soft-Shell Shrimp Classification

Jan 06, 2026arXiv.org

This study addresses the challenge of cephalothorax-abdomen separation in soft-shell shrimp post-harvest, which compromises visual quality and consumer acceptance. Traditional manual sorting suffers from low efficiency, poor consistency, and an inability to preserve freshness. To overcome these limitations, this work proposes the first application of deep learning–based image recognition for real-time automated classification of soft-shell shrimp immediately after harvest. A convolutional neural network (CNN) model, integrated with computer vision techniques, enables high-accuracy, automated quality assessment. The proposed method significantly improves classification accuracy and processing throughput while reducing reliance on manual labor, thereby minimizing handling time and better preserving product freshness. This approach advances the intelligent transformation of aquatic product processing and supports industry efforts to meet growing demand for high-quality seafood.

0 citationsRead paper

Domain Adaptation for Japanese Sentence Embeddings with Contrastive Learning based on Synthetic Sentence Generation

Mar 12, 2025

This paper addresses the challenge of domain adaptation for sentence embedding in low-resource languages—specifically Japanese—where large-scale labeled data are scarce. To this end, we propose SDJC, a framework that leverages a syntax-preserving synthetic sentence generator to produce grammatically consistent yet semantically divergent sentence pairs, enabling effective unsupervised contrastive learning for efficient backbone model adaptation. Our key contributions are threefold: (1) the first dependency-syntax-constrained synthetic sentence generation mechanism; (2) JSTS—the first Japanese Semantic Textual Similarity benchmark covering multiple domains and difficulty levels—filling a critical evaluation gap; and (3) a pipeline integrating machine translation augmentation with contrastive fine-tuning, yielding substantial improvements on downstream tasks. We publicly release the JSTS dataset, training code, and the adapted Japanese sentence embedding models.

0 citationsRead paper

Recent publications

Latest Papers

A feasibility study on filtering low-accessibility web pages considering color vision deficiency

Jun 20, 2026

This study addresses the challenge of color accessibility for users with color vision deficiencies when browsing the web by proposing a machine learning–based automatic filtering approach. It pioneers the use of predictive modeling to identify and exclude webpages that violate Color Universal Design (CUD) principles, thereby exhibiting low accessibility. The method integrates CUD guidelines into a tailored evaluation metric, and experiments on 21 real-world webpages demonstrate that the model achieves a maximum AUC of 0.76, confirming the feasibility of automatically enhancing web color accessibility. This work offers a scalable technical pathway and practical insights for improving information accessibility for individuals with color vision deficiencies.

0 citationsRead paper

Exploring the Relationship Between Local Election Results and Online Public Opinion in Taiwan: A Case Study of Taitung County

Jan 06, 2026arXiv.org

This study examines local elections in Taitung County, Taiwan, to investigate the relationship between online public sentiment and actual voting outcomes, addressing a gap in empirical research on this topic. By systematically analyzing the correlation between social media discourse volume during the election period and candidates’ vote shares, the research establishes an empirical framework linking digital舆情 to voter behavior. Findings indicate that online discussion can, to a certain extent, serve as a reliable predictor of electoral results, offering new insights and practical implications for political communication, campaign strategy optimization, and election forecasting in the digital era. The study also acknowledges limitations concerning data representativeness and methodological constraints.

0 citationsRead paper

An Empirical Study on User Profile Analysis and SEO Performance: A Case of Taiwan Cultural Memory Bank 2.0

Jan 06, 2026arXiv.org

This study aims to enhance the digital outreach effectiveness and service quality of Taiwan’s Cultural Memory Bank 2.0 platform. By integrating the Visitor Relationship Management (VRM) framework, user behavior analytics, and Search Engine Optimization (SEO) assessment, the research empirically examines user demographics, browsing patterns, and engagement behaviors while quantifying the platform’s search visibility and organic traffic performance. Key user segments and their behavioral characteristics are identified, and combined with SEO metrics to formulate data-driven strategies for website optimization and social media promotion. The primary contribution lies in the synergistic integration of user personas with SEO indicators, offering actionable pathways to improve both user experience and online visibility for digital cultural platforms.

0 citationsRead paper

Deep Learning-Based Image Recognition for Soft-Shell Shrimp Classification

Jan 06, 2026arXiv.org

This study addresses the challenge of cephalothorax-abdomen separation in soft-shell shrimp post-harvest, which compromises visual quality and consumer acceptance. Traditional manual sorting suffers from low efficiency, poor consistency, and an inability to preserve freshness. To overcome these limitations, this work proposes the first application of deep learning–based image recognition for real-time automated classification of soft-shell shrimp immediately after harvest. A convolutional neural network (CNN) model, integrated with computer vision techniques, enables high-accuracy, automated quality assessment. The proposed method significantly improves classification accuracy and processing throughput while reducing reliance on manual labor, thereby minimizing handling time and better preserving product freshness. This approach advances the intelligent transformation of aquatic product processing and supports industry efforts to meet growing demand for high-quality seafood.

0 citationsRead paper

Domain Adaptation for Japanese Sentence Embeddings with Contrastive Learning based on Synthetic Sentence Generation

Mar 12, 2025

This paper addresses the challenge of domain adaptation for sentence embedding in low-resource languages—specifically Japanese—where large-scale labeled data are scarce. To this end, we propose SDJC, a framework that leverages a syntax-preserving synthetic sentence generator to produce grammatically consistent yet semantically divergent sentence pairs, enabling effective unsupervised contrastive learning for efficient backbone model adaptation. Our key contributions are threefold: (1) the first dependency-syntax-constrained synthetic sentence generation mechanism; (2) JSTS—the first Japanese Semantic Textual Similarity benchmark covering multiple domains and difficulty levels—filling a critical evaluation gap; and (3) a pipeline integrating machine translation augmentation with contrastive fine-tuning, yielding substantial improvements on downstream tasks. We publicly release the JSTS dataset, training code, and the adapted Japanese sentence embedding models.

0 citationsRead paper