Knowledge graphs play a pivotal role in structuring human knowledge within artificial intelligence systems. Nonetheless, knowledge distribution is markedly uneven across languages, and linguistic community activity can hinder the performance and scale. Cross-lingual transfer learning emerges as a predominant effective strategy to surmount linguistic barriers, facilitating knowledge transfer across natural languages. This paper reviews cross-lingual knowledge acquisition for knowledge graphs, offering the first systematic integration of cross-lingual transfer paradigms and resources in this field. It critically examines the state of research across subtasks (including named entity recognition, relation extraction, coreference resolution and entity linking). Despite the advancements facilitated by multilingual word embeddings, pre-trained language models and large language models, persistent challenges such as language bias-induced alignment difficulties and low transfer efficiency continue to impede progress. Enhancing model effectiveness through both paradigms and resources will benefit the future construction of multilingual or minor-language knowledge graphs.

Accurate segmentation of knee cartilage and meniscus in magnetic resonance imaging (MRI) is essential for the early detection and monitoring of complications such as cartilage erosion and osteoarthritis. Yet, manual annotation remains time-consuming, subjective, and inefficient for routine clinical use. In this study, we introduced KneeXNet-2.5D, a clinically oriented and explainable deep learning framework for accurate and efficient knee cartilage and meniscus segmentation in sagittal MRIs. Unlike traditional 3D segmentation methods, the proposed model employs a 2.5D architecture to capture inter-slice spatial context, achieving high segmentation accuracy while maintaining computational efficiency and optimal resource utilization. We further incorporated targeted image augmentation, including synthetic noise injection, to enhance the AI model robustness against medical imaging variability. The efficient design of the 2.5D model allows for reduced resource consumption, making it suitable for deployment in healthcare settings with limited computational infrastructure, particularly in low-resource hospitals and rural care environments. To enable open scientific research and ensure reproducibility, we constructed a gold-standard, manually segmented knee MRI dataset and publicly released it alongside the annotation guideline, source code, trained AI models, and a lightweight software application. An entropy-based AI explainability strategy was developed to highlight high-uncertainty regions that are most influential to model predictions, advancing transparency and interpretability. Clinical relevance and anatomical validity were further assessed through expert review by board-certified orthopedic surgeons. Together, these contributions demonstrate the AI model’s anatomical fidelity, interpretability, and readiness for integration into musculoskeletal imaging workflows.

Megaprojects, as vital instruments of societal and infrastructural transformation, are often challenged by cost overruns, delays, and evolving risks—issues that can compromise both their immediate success and long-term sustainability. Traditional risk management approaches, grounded in static and reactive frameworks, fall short in addressing the complexity and dynamism of such large-scale initiatives. This position paperproposes an Interactive and Adaptive AI-based Decision SupportSystem (DSS) that integrates Interactive Machine Learning(IML) and Adaptive Deep Learning with domain expert input. Central to this approach is the concept of hybrid intelligence, where AI-driven analytics and human judgment are combined synergistically to enhance decision-making. The system embedsexpert-in-the-loop mechanisms to enable continuous, contextaware learning from both historical data and real-time feedback. This produces interpretable, real-time insights for proactive risk identification, assessment, and forecasting. The hybrid intelligence framework promotes greater resilience, adaptability, andtransparency—contributing to more sustainable, informed, andethically responsible governance of complex megaprojects.

Large language models (LLMs) are reshaping education by altering how knowledge is accessed, interpreted, and constructed. This paper reviews 98 academic sources-including peer-reviewed articles, conference papers, and expert reports-to examine the pedagogical and institutional implications of LLM integration. We argue for a cautious but proactive redesign of education grounded in critical AI literacy, ethical safeguards, and task authenticity. LLMs are framed as intelligent interfaces that require new approaches to assessment, instruction, and learning design. The paper proposes actionable principles for AI-integrated education and calls for longitudinal research on the cognitive and developmental impact of scaffolded LLM use in adolescence.

In emergency situations, accurately estimating an individual’s time of escape can greatly improve safety and support more informed decision-making. This paper presents a real-time framework for the Estimation Time of Escape (ETE) in indoor environments using computer vision and Internet of Things (IoT) technologies. Unlike prior approaches, this work emphasizes real-time applicability. By also including unobservable zones between observed areas, the complexity of the framework increases while also making it more robust to different scenarios and building structures. The ETE framework utilizes IP cameras in conjunction with a YOLOv11 object detection model and the ByteTrack tracking algorithm to detect and track individuals in multiple rooms. A homography-based perspective transformation is applied to map the tracked positions from the video to real-world positions on the floor. This enables accurate estimation of speed and direction. The framework operates in real-time and dynamically adapts to changing conditions, such as variations in walking speed or direction. It can also predict potential congestion near exits by analyzing comparative escape times. The framework is evaluated using multi-person scenarios, achieving a mean absolute error of up to 0.83s for two-person scenarios and an average of under 1.5s overall. The results show the framework’s robustness and real-world applicability for real-time evacuation monitoring. This framework offers a scalable and adaptable solution with potential applications in public safety, building evacuation systems, and smart city infrastructure.

Understanding travel behaviour is crucial for developing inclusive, adaptable transportation systems in ageing cities. Yet many public authorities lack the expertise and budget for advanced analytics. We present a modular, multi-agent framework that employs Large Language Models (LLMs) to lower the entry barrier to data-driven mobility analysis. The pipeline extracts interpretable, policy-oriented insights from unstructured survey data and outputs structured reports without manual coding. We demonstrate the approach using Malmö's 2023 travel survey for residents aged 65+, and benchmark its recommendations against those of (i) a single-agent LLM service (Single Agent), representative of current commercial offerings, and (ii) a human expert baseline (Human), both evaluated in a blinded expert review. The system achieves superior reasoning quality compared to the single-agent baseline, while performing slightly lower on interpretability and focus, and approaches expert-level quality overall - at a fraction of the cost and effort. Key constraints are the current lack of real-time data ingestion and dependence on proprietary commercial APIs. The study provides an open-source proof of concept showing how multi-agent LLMs can make urban mobility analytics more accessible, transparent, and timely. All code and evaluation materials are publicly available.

Occupancy estimation in smart buildings is essential for optimizing resource usage and enhancing operational efficiency. Existing estimation methods predominantly rely on cameras or advanced sensor fusion techniques, which, while accurate, are often expensive, invasive, and raise privacy concerns. Additionally, these approaches frequently require extra hardware, increasing installation complexity and operational costs. A significant gap in the literature lies in the limited use of existing smart building infrastructure, such as detection systems and booking data, for people counting. This study addresses these limitations by exclusively utilizing two binary PIR sensors (in-door and in-room) and booking data. Since PIR sensors and booking systems are already integrated into most smart building infrastructures, leveraging these existing resources helps reduce costs and simplifies implementation. The primary goal is to estimate the number of people between each in-door sensor trigger using machine learning models by incorporating people counting levels and time thresholds. Among the evaluated machine learning algorithms, the Extra Trees Classifier delivered strong performance, achieving 68.5% accuracy when the estimated occupancy differed from the actual count by at most one person, and 81.56% with a tolerance of two. These results are based on periods when the room was occupied. When both occupied and unoccupied periods were included, the accuracy was 96.10% for ±1 tolerance. Moreover, incorporating booking data enhanced people counting accuracy by 4%. The study also explores the method's ability to identify underutilization and overutilization by comparing estimated occupancy with booking records and seating capacity, thereby supporting enhanced space management in smart buildings.

People counting in smart buildings is crucial for the efficient management of building systems such as energy, space allocation, efficiency, and occupant comfort. This study investigates the use of two non-invasive binary Passive Infrared (PIR) sensors for estimating the number of people in seven office rooms with different people counting intervals. Previous studies often relied on sensor fusion or more complex signal-based PIR sensors, which increased hardware costs, raised privacy concerns, and added installation complexity. Our approach addresses these limitations by utilizing fewer sensors, reducing hardware costs, and simplifying installation, making it scalable and flexible for different room configurations, while also ensuring high consideration of privacy. Additionally, binary PIR sensors are typically part of smart building systems, eliminating the need for additional sensors. We employed several machine learning methods to analyze motion detected by binary PIR sensors, imp roving the accuracy of people counting estimates. We analyzed important features by extracting event count, duration, and density from sensor data, along with features from the room’s shape, to estimate the number of people. We used different machine learning models for estimating the number of people. Models like Gradient Boosting, XGBoost, MLP, and LGBM demonstrated superior performance for their strong ability to handle complex, non-linear relationships in sensor data, high-dimensional datasets, and imbalanced data, which are common challenges in people counting tasks using PIR sensors. These models were evaluated using performance metrics such as accuracy and F1-score. Additionally, the results show that features such as passage events and the number of detected events, combined with machine learning algorithms, can achieve good accuracy and reliability in people counting.

The state space explosion, a challenge analogous to that encountered in a Petri net (PN), has constrained the extensive study of fuzzy Petri nets (FPNs). Current reasoning algorithms employing FPNs, which operate through forward, backward, and bidirectional mechanisms, are examined. These algorithms streamline the inference process by eliminating irrelevant components of the FPN. However, as the scale of the FPN grows, the complexity of these algorithms escalates sharply, posing a significant challenge for practical applications. To address the state explosion issue, this work introduces a parallel bidirectional reasoning algorithm for an FPN that utilizes reverse and decomposition strategies to optimize the implementation process. The algorithm involves hierarchically dividing a large-scale FPN into two sub-FPNs, followed by a converse operation to generate the reversal sub-FPN for the right-sub-FPN. The detailed mapping between the original and reversed FPNs is thoroughly discussed. Parallel reasoning operations are then conducted on the left-sub-FPN and the resulting reversal right-sub-FPN, with the final result derived by computing the Euclidean distance between the outcomes from the output places of the two sub-FPNs. A case study is presented to illustrate the implementation process, demonstrating the algorithm's significant enhancement of inference efficiency and substantial reduction in execution time.

基于可逆和动态分解机制的层次化FPN并行故障诊断

与Petri网类似,模糊Petri网(fuzzy Petri net, FPN)的研究同样受到状态空间爆炸问题的限制。目前,基于FPN的推理算法主要依赖于正向、反向和双向等机制。这些算法通过消除FPN中不相关的部分来简化推理过程。然而,随着规模的扩大,基于FPN的相关应用算法的复杂度迅速增加,这给基于FPN的推理算法的实际应用带来重大挑战。为解决状态爆炸问题,本文提出一种基于可逆和动态分解机制的FPN双向推理算法,以优化推理过程。该算法将层次化后的FPN分解为左右两个子网;然后,深入分析FPN原网与其逆网元素之间的对应关系,提出FPN逆网生成算法,用于生成右子网的逆网;最后,在左子网与右子网的逆网上同时执行推理算法,通过计算两子网输出位置之间的欧式距离得到最终结果。案例表明,本文提出的推理算法显著提高了推理效率,大幅缩短了执行时间。

We evaluate deep learning architectures for rat pose estimation using a six-camera system, focusing on ResNet and EfficientNet across various depths and augmentation techniques. Among the configurations tested, ResNet 152 with default augmentation provided the best performance when employing a multi-perspective network approach in the controlled experimental setup. It reached a Root Mean Squared Error (RMSE) of 8.74, 8.78, and 9.72 pixels for the different angles. The utilization of data augmentation revealed that less altering yields better performance. We propose potential areas for future research, including further refinement of model configurations, more in-depth investigation of inference speeds, and the possibility of transferring network weights to study other species, such as mice. The findings underscore the potential for deep learning solutions to advance preclinical research in behavioral neuroscience. We suggest building on this research to introduce behavioral recogniti on based on a 3D movement reconstruction, particularly emphasizing the motoric aspects of neurodegenerative diseases. This will allow for the correlation of observable behaviors with neuronal activity, contributing to a better understanding of the brain and aiding in developing new therapeutic strategies.