The objective of this study is to evaluate the ability of large language models (LLMs) to generate an agreement based on social welfare in multi-issue negotiation. In the experiment, prompts were designed that in-cluded the negotiation theme, issues, options within each issue, weights of the issues, utility values of the options, the method for calculating social welfare, and the output format. These prompts were then input into each LLM. All negotiations described in the prompts were two-party negotiations, and two types of negotiation scenarios were considered: one where the agents’ utilities were aligned, and one where they were conflict-ing. For evaluation, the generated agreements and their corresponding social welfare values were classified into five categories. As a result, GPT-4o-mini and Llama 3.3 70B showed strong performance in scenarios with aligned interests, while significant performance differences were observed among models in scenarios with conflicting interests. These results sug-gest that while LLMs can handle simple negotiation scenarios, they face challenges in more complex ones. Future work includes evaluating the consistency of the generation process, improving computational accuracy, and diversifying negotiation scenarios.

抄録全体を表示

This paper presents a novel approach to tourism information recommendation that incorporates cultural context analysis based on multilingual website access patterns. We analyzed two years of access data from Kagoshima Prefecture's multilingual tourism website across six languages, revealing distinct behavioral characteristics for each language group. Based on these findings, we propose enhancements to our existing eye-tracking-based recommendation system by integrating language-specific weighting in TF-IDF processing and cultural filtering mechanisms. The proposed system aims to provide culturally-appropriate recommendations while encouraging discovery of new tourism experiences.

抄録全体を表示

In our research, we are developing a keyword recommendation system for tourism-related search terms by extracting user preferences from gazing sentences and still images obtained based on the user's gaze information while browsing websites. However, many Web sightseeing sites contain not only text and still images, but also “explanatory videos” of sightseeing spots. Therefore, we propose the addition of a function for extracting user preferences from content containing gazing videos to existing system.

抄録全体を表示

Duality of codes over a commutative non-unitary ring of order 9 using the symplectic inner product is explored. For codes over this ring, there are two associated ternary codes, called residue code and torsion code. By studying the residue and torsion codes, conditions for symplectic orthogonality will be explored and a mass formula for symplectic self-orthogonal codes will be derived.

抄録全体を表示

Self-dual codes have been widely studied in cod-ing theory, both over nite elds and over rings. Af-ter the discovery that several well-known families of non-linear binary codes are intimately related to lin-ear codes over the ring Z4 (the ring of integers mod-ulo 4), much attention have been given to codes over rings, specically rings with unity. Only recently re-searchers begun to explore non-unital rings as alpha-bet [1, 2] The absence of a multiplicative identity cre-ates difficulties in dealing with the usual concept of self-duality in a code. As a result, the notion of quasi self-dual (QSD) codes was introduced [1], that is, self-orthogonal codes of size pⁿ. The goal of this paper is to present some basic theory on the duality of codes over a non-local, non-unital ring Hp in the classication of Fine [3]. We characterize self-orthogonal, self-dual and quasi self-dual codes over Hp.

抄録全体を表示

Encryption is used as a means of ensuring the confidentiality of information communications over the Internet, and many of these methods use a hybrid method of secret key cryptography and public key cryptography. Secret key cryptography is suitable for encrypting and transferring large amounts of data, but the key must be distributed in advance. In our information and communications network society, where we communicate with an unspecified number of people, the problem of key distribution is solved by public key cryptography. The security of public key cryptography currently in practical use relies on the difficulty of efficiently solving the prime factorization problem of large numbers and the discrete logarithm problem on elliptic curves. However, with the advent of practical quantum computers, it has been shown that these problems may be solved efficiently. Of course, secure data transfer methods for quantum communication have also been proposed, but it is also important to develop encryption methods using current communication means that cannot be decrypted within a practical time frame even by quantum computers. This type of encryption scheme is collectively called PQCC, and several schemes have been proposed. Here, we propose a scheme that uses a self-dual code over the integer quotient ring.

抄録全体を表示

The Erdős-Straus Conjecture (ESC) states that for all integers n greater than or equal to two, the number four divided by n is a sum of three positive unit fractions. Although computationally verified for all integers up to ten to the seventeenth power, no general constructive proof has been available, particularly for prime numbers and certain notorious modular-exception residues. This paper presents a comprehensive theorem, based on our divisibility condition, that provides a novel constructive approach with explicit solution formulas. Remarkably, this divisibility condition also functions as a necessary and sufficient condition for the solvability of the ESC, including all prime cases and previously known Modell exceptional values. Our resulting algorithm achieves a success rate exceeding ninety-two percent for massive numbers, including primes exceeding three thousand digits, and obtains a one hundred percent success rate for the tested class of integers involving all Mordell exceptional values, with solution times under one hundred and forty-two milliseconds per case on consumer hardware. We demonstrate that discovering mathematical structure can overcome computational intractability more effectively than raw computing power, achieving a performance improvement of an astronomical scale over brute force methods. These results have fundamental implications for cryptographic security, suggesting that, complementing the preparation for advances in quantum computing, it is also crucial to explore the potential for hidden mathematical structures in cryptographic problems.

抄録全体を表示

In recent years, data science education has been actively advocated around the world. Much of the education have involved deepening information processing education based on statistics and programming, which still gives the impression of being a high hurdle for non-STEM students. In Japan, it is recognized that nurturing human resources with knowledge of data science, regardless of their major in science, humanities, or social science, is important for the development of the country, and so it has been adopted as a national policy. In this paper, we look back on the history of ICT and data science education encouraged by the Japanese government, and examine its effectiveness. Then Some of examples and discussion of the state of data science education in several universities and junior colleges are provided.

抄録全体を表示

As data science becomes vital for solving complex societal challenges, inclusive data literacy is needed across all disciplines. However, in Japan, rigid divisions between STEM and non-STEM fields limit access for students in humanities, social sciences, and sports. This paper explores the systemic, cultural, and educational barriers facing non-STEM students in acquiring data skills. Drawing on policy trends and institutional practices, we identify key challenges such as narrow curricula, negative attitudes toward quantitative subjects, and unclear career relevance. We advocate for interdisciplinary, context-based learning that empowers diverse learners to apply data science for social good. This first paper in a two-part series highlights the need for inclusive education reforms to build a more equitable digital society.

抄録全体を表示

This study examined the impact of a data science lecture for social inclusion on 168 STEM and non-STEM undergraduates (89 in the Sports and 79 in the Regular program) at Rissho University. Using a mixed-methods approach (survey analysis and NLP-driven qualitative text analysis), it explored evolving student perceptions. Quantitative results showed increased understanding and interest. Qualitative analysis revealed that despite initially lower theoretical confidence, non-STEM students creatively proposed practical, socially oriented applications. Key themes included "Data Utilization," "Social Issues," and "Disability and Welfare," with a notable shift from abstract concepts to concrete problem-solving. Findings suggest that interdisciplinary, context-based pedagogy emphasizing real-world problems and social impact enhances data literacy and fosters inclusive innovation.

抄録全体を表示

In fast-paced development scenarios—such as Game Jams or prototype builds for exhibitions—designers often lack the time to systematically check whether players fully understand the in-game tutorials, whether the level pacing feels balanced, or whether the overall play experience is satisfying. This paper presents an evaluation and optimization approach powered by large language models (using ChatGPT as an example) to help designers during the prototype phase. Our method identifies missing tutorial elements, predicts where players are likely to get stuck, estimates completion times, and generates targeted improvement suggestions. We validate its feasibility and effectiveness by applying it to a nonlinear sandbox level in a Souls-like game. Finally, we compare AI-driven evaluation with traditional manual assessment in terms of accuracy, efficiency, and objectivity.

抄録全体を表示

In short-term game development projects such as Game Jams, student teams often face challenges in balancing creativity with implementation due to strict time limits and high coordination costs. This study investigates the use of large language models (LLMs) within a modular development workflow for student game projects developed in Unity. We begin by introducing the principles of modular development and LLMs, then categorize mainstream tools into three groups—programming assistance, AI agents, and conversational AI—analyzing their respective strengths and appropriate use cases. The discussion then shifts to three core subsystems—core gameplay logic, user interface (UI), and tools & pipeline automation—evaluating each in terms of code generation, cross-module collaboration, and validation. Finally, we present practical guidelines, including context provision, iterative decomposition, tool chaining, and prompt optimization, while also addressing limitations such as performance tuning and closed-source debugging. Our findings suggest that LLMs can substantially shorten development cycles for well-defined, template-based modules, but still require human oversight in performance-sensitive tasks and contexts requiring global consistency. This research provides a reference framework for both educational practice and the design of future AI-driven game development platforms.

抄録全体を表示

Generative AI has increasingly become an essential tool in game art creation; however, significant gaps persist between students and professional artists in workflow practices, prompt design, and image control. This exploratory study conducts a comparative analysis between case examples drawn from student experiments and publicly documented professional workflows. Results reveal that students typically rely on intuitive trial and error prompting often yielding inconsistent or sub-optimal outputs whereas professionals employ structured prompt engineering and systematic refinement to ensure image quality and stylistic coherence. In response, we propose a lightweight educational framework comprising a Structured Prompting Template and an Iterative Generation Cycle scaffold. Preliminary simulation exercises demonstrate that these tools significantly enhance students’ prompt coherence, visual consistency, and critical evaluative engagement. This paper provides foundational guidance for integrating generative AI into game art education and suggests avenues for future pedagogical development across diverse visual domains.

抄録全体を表示

This study introduces an efficient approach to constructing 3D game environments in educational contexts, drawing on the authors’ practical experience in developing an action role-playing game (ARPG) with the Unity engine. By examining the workflows of small teams working under tight budgetary and time constraints, the paper underscores the role of modular design and asset management in creating environments that are both stylistically coherent and resource-efficient. It further highlights the pedagogical value of asset reuse and recombination, demonstrating how these practices can be incorporated into game design education. Beyond technical considerations, the study emphasizes that level design should be viewed not as a one-time creative act, but as a collaborative and iterative process shaped through communication and feedback. The proposed workflow therefore serves both as practical guidance for developers and as a structured framework for teaching environment design in game education.

抄録全体を表示

In recent years, many companies have had problems with information security. These include data leaks, phishing email etc. which may cause financial loss, reputational damage, productivity decline, safety/health damage and so on. Even though organizational leaders are more aware of these problems now, many still don’t fully understand the current information security status of their organization. The visualization of metrics resulting from security assessments process will help them to identify issues accurately. We believe that KRIs based analysis is necessary to create metrics that appear to be effective, and the importance of each KRI varies depending on the characteristics of the organization. In this paper we propose a method for extracting KRIs that reflects the characteristics of an organization type and provide examples.

抄録全体を表示

This study theoretically examines the significance of Responsible AI as a framework for assessing the environmental and ethical aspects of AI technology within ESG investments. RAI not only ensures principles such as transparency, fairness, accountability, and safety in the design, development, and operation of AI but also serves as a means to assess and visualize its environmental impact, including greenhouse gas emissions and electricity consumption. By focusing on the intersection of RAI and ESG investment, this paper discusses how the implementation of RAI can enhance the accuracy of investors’ decision-making while promoting companies’ intrinsic sustainability strategies.

抄録全体を表示

In this study, we report on the summarization of natural language questionnaires using generative AI in order to solve this problem in summarizing the results of free-form questionnaires. In particular, generative AI excels in its ability to extract key points from a large amount of text data and summarize them in a concise and easy-to-understand manner, and is considered to be able to contribute to the efficient and objective analysis of questionnaire data.

抄録全体を表示

This study was conducted to clarify difficulties that students with prior experience using rubrics face when tackling an assignment without one. After a questionnaire survey was administered to 138 female junior college students in Japan, the collected data were analyzed using exploratory factor analysis and multiple regression analysis. The results identified two factors constituting these difficulties: a "Psychological Barrier to Asking Questions" and "Difficulty in Task Execution." Furthermore, multiple regression analysis revealed that studstudents who rated the usefulness of rubrics higher anticipated significantly greater difficulty in task execution ( p < 0.001) . Results suggests that rubrics function as a "learning compass" for the student learning process, highlighting the need for alternative support that addresses both procedural guidance and students′ psychological barriers when rubrics are not provided.

抄録全体を表示

[Purpose] The COVID-19 pandemic made it necessary to provide first-year nursing students with remote access to lectures on the structure and function of the human body (SFHB) via Google Classroom (GCR). However, I had doubts about whether these remote lectures would be effective in teaching three-dimensional anatomy, and I considered introducing virtual reality (VR) and augmented reality (AR) teaching materials. Furthermore, with the aim of creating a rubric for VR/AR teaching materials, I conducted preliminary effect measurements and reported on them. [Method] Seventy-seven students were asked to give their impressions in an anonymized questionnaire regarding the use of VR/AR teaching materials similar to the videos used in standard SFHB classes on line, and 38 responded (valid response rate, 49.4%). The evaluation was based on a comparison between the VR/AR lectures and face-to-face lectures using videos of the responses to: questionnaires about the teaching materials and to class improvement questionnaires; the student feedback after each lecture; and the levels attained in examinations. The rubric incorporated evaluation scales, evaluation viewpoints, and evaluation criteria (Table,1). The university is a member of the Compensation System for Public Transmission for Educational Purposes; this research was approved by my former university. [Results] 1. Similar to the evaluations of conventional videos, 86.8% of the evaluations of VR/AR teaching materials were reported by students as “very interesting” or “interesting.” 2. When students were asked if they would like to use these materials in future lectures and preparations for national exams, 52.6% said they would, and 5.3% said the materials were unnecessary. 3. In the questionnaire, there were positive responses such as “because it is easy to understand” and “VR is easier to remember.” In contrast, some students expressed opinions such as “Printed material is enough.” The scores in the class improvement questionnaire were 2.95–3.00 points out of 3 for all items. In the rubric created from the above, the students were considered to have achieved “acceptance of VR/AR,” “acquisition of knowledge,” and “three-dimensional understanding,” but “engagement” scored only 52.6%. Students found “filling in blank body maps” difficult. In addition, when I asked about the potential use of XR (extended reality) teaching materials for subjects other than SFHB, pathology was the most popular, followed by maternity, geriatrics, and physical assessment. [Discussion] The virtual space service called the metaverse is expected to increase in popularity. In this survey, students were interested in VR/AR, but because the teaching materials used were not necessarily designed for nurses the “engagement” score was 52.6%. According to a survey by the Japanese Nursing Association, 9.2% of students are offered VR/AR in their lectures—a percentage that is still small. I therefore decided to use VR/AR teaching materials for nursing training to improve engagement.

抄録全体を表示

The COVID-19 pandemic has significantly accelerated the digitization of daily life, including the widespread adoption of teleworking and online education. In the field of counseling, remote modalities have become increasingly prevalent as a means of minimizing in-person contact, with online platforms now commonly utilized for individual consultations. This digital shift has also extended to group therapy and self-help groups, prompting active discourse on both the benefits and limitations of virtual formats. This study explores the advantages and challenges associated with online self-help group activities, using the case of “Club Minashigo (Orphan) Gathering for the Truly Alone,” which has operated via Zoom since its inception in 2022. The group’s online structure exemplifies the potential of digital platforms to facilitate participation across geographic boundaries and to foster social connection among individuals experiencing isolation. Findings indicate that the elimination of spatial and financial barriers through online meetings lowers the threshold for engagement, thereby enhancing accessibility. However, the study also identifies several areas requiring further investigation: the difficulty of evaluating therapeutic effectiveness, the limitations in cultivating trust and interpersonal relationships compared to face-to-face interactions, and the complex implications of anonymity within virtual settings.

抄録全体を表示

Currently, there is a problem with the lack of doctors who diagnose colorectal cancer. Especially in Japan, colorectal cancer patients are increasing with the aging society. Therefore, the construction of automatic diagnosis support system that substitutes double and triple check in colorectal cancer cytology will be indispensable. In this study, we constructed a system to automatically identify colorectal cancer, adenoma and normal cells by changing the layer structure based on Convolutional Neural Network and Fully Convolutional Network. As a result, the maximum value of classification accuracy is 95.22% by VGG16 based Convolutional Neural Network and 96.52% for U-Net based Fully Convolutional Network. The result shows that this system opens up the possibility of the system which substitutes double check and triple check of colorectal cancer cytology.

抄録全体を表示

This study develops and evaluates two Retrieval-Augmented Generation (RAG)-based chatbot systems for municipal information services, offering deployment guidelines for municipalities with different technical capacities. The first system, built with LangChain, allows advanced code-level customization and flexibility for complex configurations and integration with existing infrastructure. The second system, constructed with Dify, is a GUI-based platform that supports rapid deployment and low-code operation. Both systems were equipped with key security features, including detection of personally identifiable information (PII), prevention of prompt injection, and request rate limiting. Evaluation was carried out using input data modeled after real administrative inquiries in simulated scenarios. Findings suggest that LangChain is more effective where strict oversight and detailed management are required, while Dify, despite limited code-level flexibility, excels in speed and ease of maintenance. This makes Dify a practical option for municipalities with limited technical expertise.

抄録全体を表示

This paper mainly explores the application of visual media in alleviating family conflicts and its educational psychological significance. With the widespread presence of family conflicts in modern society, issues such as intergenerational conflicts and emotional alienation among family members have emerged. By integrating principles of educational psychology, the study analyzes how visual media, especially animated works, help family members identify and regulate emotions through visual language and emotional resonance, thus promoting communication and emotional understanding in parent-child relationships. The paper also discusses the educational functions of visual media and its importance in family education, supported by theoretical insights.

抄録全体を表示

In this study, we propose a novel approach, "Supervised Generative Forming," which generates and manipulates images while reflecting the creator's intent through the integration of interactive artificial intelligence with image recognition technology. In conventional generative art, deriving rules from the desired final shape has posed a significant challenge. However, this system provides visual feedback at each stage of rule application. This approach enables the rules to be adjusted incrementally to achieve a gradual convergence between the initial state and the desired final shape. This approach facilitates the development of sophisticated models that preserve the organic shapes observed in nature while enabling creators to maintain oversight during the design process.

抄録全体を表示

Most existing fatigue detection systems are trained on facial datasets primarily composed of Western individuals, which limits their ability to capture the unique characteristics of Asian faces, particularly in terms of eye size and shape. Key physiological indicators such as blinking frequency and eye closure duration have been shown to strongly correlate with fatigue levels. However, current datasets may lead to insufficient accuracy for fatigue detection in Asian populations. This study examines publicly available facial image datasets to evaluate their suitability for Asian fatigue detection. Furthermore, we propose the construction of a specialized dataset focused on blinking features specific to Asian individuals.

抄録全体を表示

This study aims to clarify the educational effects of PBL courses at our university. In this paper, we compare questionnaires administered to students at the time of assignment and after course completion to examine changes in expectations and satisfaction levels. Significant differences are identified in 21 of 59 items, with notable improvements in evaluations of theme difficulty, knowledge and skill acquisition, and intergenerational collaboration. These findings provide implications for improving educational design, highlighting directions to better meet students’ expectations and enhance satisfaction with learning activities.

抄録全体を表示