Seung-Hyun Han

Large language models (LLMs) show limited capability in processing low-resource historical languages due to insufficient training data and domain-specific linguistic structures. Korean Literary Sinitic (KLS), the principal written medium of the Joseon dynasty, remains particularly under-resourced despite its lexical overlap with modern Korean and shared script with classical Chinese. To enable systematic evaluation in this domain, we introduce KLSBench, a comprehensive benchmark for assessing LLM performance on KLS. KLSBench contains 7871 instances sourced from Joseon dynasty civil service examination archives and parallel corpora of the Four Books, and encompasses five task categories: classification, retrieval, punctuation restoration, natural language inference, and translation. Our evaluation suggests KLSBench could work as an effective diagnostic tool that distinguishes lexical recall from deeper linguistic comprehension in low-resource historical languages. Beyond establishing evaluation baselines, KLSBench provides practical frameworks for deploying LLM-based tools in digital humanities contexts, including automated annotation systems and intelligent search interfaces for classical text repositories.

In recent years, significant advancements have been made in spatiotemporal sequence prediction, with PredRNN++ emerging as a powerful model due to its superior ability to capture complex temporal dependencies. However, the current unidirectional nature of PredRNN++ limits its ability to fully exploit the temporal information inherent in many real-world sequences. In this research, we propose an enhancement to the PredRNN++ model by incorporating a bidirectional network, enabling the model to consider both past and future contexts during prediction. This bidirectional extension enhances the model’s ability to predict sequences accurately and reliably, especially for data with intricate temporal patterns. Our experimental results demonstrate that the Bidirectional PredRNN++ outperforms the original model across several benchmark datasets, highlighting its potential for a wide range of applications in spatiotemporal data analysis.

In recent years, significant advancements have been made in spatiotemporal sequence prediction, with PredRNN++ emerging as a powerful model due to its superior ability to capture complex temporal dependencies. However, the current unidirectional nature of PredRNN++ limits its ability to fully exploit the temporal information inherent in many real-world sequences. In this research, we propose an enhancement to the PredRNN++ model by incorporating a bi-directional network, enabling the model to consider both past and future context during prediction. This bi-directional extension aims to improve the predictive accuracy and robustness of the model, particularly in tasks involving complex temporal dynamics. Our experimental results demonstrate that the bi-directional PredRNN++ outperforms the original model across several benchmark datasets, highlighting its potential for a wide range of applications in spatiotemporal data analysis.

With the rapid expansion of the virtual reality (VR) market, user interest in VR games has increased significantly. However, empirical research on the user experience in VR games remains relatively underdeveloped. Despite the growing popularity and commercial success of VR gaming, there is a lack of comprehensive studies analyzing the impact of different aspects of VR games on user satisfaction and engagement. This gap includes insufficient research on the categorization of VR game genres, the identification of user challenges, and variations in user experiences across these genres. Our study aims to fill this gap by analyzing data from the Meta Quest store using K-means clustering and LDA (Latent Dirichlet Allocation) to categorize the representative genres of VR games. By employing text-mining techniques to conduct a detailed analysis of user experience, we effectively elucidate the primary issues and nuanced differences in user responses across various genres. Our findings serve as a valuable reference for researchers aiming to design games that align with VR user expectations. Furthermore, our study provides a foundational dataset for researchers aiming to enhance the user experience in VR games and suggests ways to increase the immersion and enjoyment of VR gameplay.

As virtual reality (VR) technology advances, research has focused on enhancing VR content for a more realistic user experience. Traditional emotion analysis relies on surveys, but they suffer from delayed responses and decreased immersion, leading to distorted results. To overcome these limitations, we propose an emotion analysis method using sensor data in the VR environment. Our approach can take advantage of the user’s immediate response and not reduce immersion. Linear regression, classification analysis, and tree-based methods were applied to electrocardiogram and galvanic skin response (GSR) sensor data to measure valence and arousal values. We introduced a novel emotional dimension model by analyzing correlations between emotions and the valence and arousal values. Experimental results demonstrated the highest accuracy of 77% and 92.3% for valence and arousal prediction, respectively, using GSR sensor data. Furthermore, an accuracy of 80.25% was achieved in predicting valence and arousal using nine emotions. Our proposed model improves VR content through more accurate emotion analysis in a VR environment, which can be useful for targeting customers in various industries, such as marketing, gaming, education, and healthcare.