The inherent uncertainty of concrete dam systems, together with the complex influence of environmental loads and measurement noise, makes it difficult for traditional deterministic point prediction models to provide reliable deformation forecasts. In particular, the prediction performance of conventional models is highly dependent on parameter settings, while the uncertainty and potential deviation of future displacement responses are often not fully quantified. To address these limitations, this study proposes an intelligent data-driven prediction framework for dam displacement based on the integration of convolutional neural networks and long short-term memory networks. In the proposed framework, convolutional neural networks are used to extract local feature information from monitoring data, while long short-term memory networks are employed to capture temporal dependencies in displacement sequences. The Black-winged Kite Algorithm is introduced to optimize the key parameters of the integrated multi-level network, thereby improving the accuracy and robustness of point prediction. Furthermore, quantile regression is embedded into the optimized learning framework to construct an interval prediction model for dam deformation, enabling the conditional predictive uncertainty associated with displacement evolution to be quantitatively characterized. The engineering case study and comparative analyses with other models demonstrate that the proposed model achieves improved prediction performance for the investigated monitoring point. The interval prediction results further show that, for the investigated dam and monitoring point, the proposed framework can effectively characterize conditional predictive uncertainty and provide additional information for deformation interpretation and safety assessment. Further studies involving additional monitoring points and dams are required to assess its broader applicability.
Visual analytics combines computational analysis with interactive visual representations and may strengthen pharmacovigilance, but its use in drug safety research has not been systematically characterized. This scoping review mapped the extent, characteristics, and applications of visualization and visual analytics in pharmacovigilance and identified evidence gaps. Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidance, we searched PubMed, IEEE Xplore, Web of Science, Scopus, and Google Scholar from inception through April 2026. Two reviewers independently selected studies and extracted data, which were synthesized qualitatively using predefined frameworks covering pharmacovigilance activities, analytical methods, visual representations, and interaction capabilities. Sixty-seven studies met the eligibility criteria: 41 system design studies and 26 non-system studies. Of the designed systems, 29 were classified as visual analytics systems and 12 as interactive visualization systems. Signal management was the most frequently supported activity; plots and charts were the predominant visual representations; and selecting, filtering, and drilling were the most common interactions. Only 16 systems reported formal evaluations. Visual analytics has been applied across diverse pharmacovigilance activities, particularly signal management, but the field remains limited by incomplete end-to-end workflow support, restricted analytical transparency, limited integration of complementary evidence sources, and infrequent formal evaluation of designed systems.
Agentic AI is changing enterprise cybersecurity as AI systems move beyond passive content generation toward autonomous planning, tool use, delegated execution, and operational action. As agents connect to email, code repositories, security operations center (SOC) platforms, finance workflows, cloud services, and enterprise application programming interfaces (APIs), they increasingly function as dynamic non-human identities rather than conventional software tools or service accounts. Existing identity and access management (IAM), Zero Trust, machine identity, and AI-governance approaches remain fragmented in their treatment of delegated authority, task intent, autonomy, runtime tool use, and auditable organizational consequences. This paper addresses these gaps by proposing the AIGATE (Agentic Identity Governance, Authority, Tool-Control and Evidence) Framework. AIGATE integrates eight governance layers: agent identity registration, lifecycle governance, delegated authority mapping, intent-bound access, least agency and least privilege, runtime tool-call control, audit evidence and accountability, and revocation and resilience. The framework treats agents as governed non-human enterprise identities whose actions remain attributable to designated human and organizational roles. AIGATE is developed through a structured critical synthesis of the recent literature on agentic AI security, machine identity, Zero Trust, runtime enforcement, and AI governance, with literature-derived governance requirements mapped explicitly to the eight framework layers. Three SOC, DevOps, and finance scenarios are used as illustrative applications rather than empirical validation. The contribution is an integrated governance architecture connecting identity, delegated authority, autonomy, runtime enforcement, evidence, and revocation across the agent lifecycle.
Adaptive and explainable systems are two critical concepts in personalized learning that enhance the customized experience based on the adjusted content according to the individual learning of a learner. In this paper, we present a modular and interpretable architecture for the development of a System for Adaptive and Explainable Feedback, which is a potential solution for the above-mentioned issues in personalized learning environments. The adaptive feedback provided by the System for Adaptive and Explainable Feedback is in real time. Still, even more importantly, it is justified clearly and transparently so that both the learner and instructor understand the why behind the recommendations for action given by the system. It consists of modular technologies for data collection (word and phrase occurrence, time tracking), analysis (latency, process mining), feedback generation (adaptive after-action review), and result presentation, making a modular, flexible, and scalable system to suit many educational scenarios. To evaluate the system’s functional performance, the SAEF pipeline was applied to a dataset derived from the ASSISTments platform, a well-established educational dataset widely used in learning analytics research. On a cohort of 500 student profiles, the SAEF achieved an overall recommendation accuracy of 83.6%, a weighted F1-score of 82.9%, and a mean system response time of 42.1 ms, demonstrating both the internal computational consistency and efficiency of the adaptive pipeline. These results indicate strong agreement with score-derived difficulty categories and support the internal computational consistency of the recommendation pipeline as a proof-of-concept system, though they do not constitute independent evidence of instructional appropriateness. The SAEF is designed to support learner engagement, personalize learning pathways, and foster transparency in AI-driven educational environments; a full empirical evaluation involving real-world deployment is identified as the primary direction for future work. The perceived understandability, trustworthiness, and pedagogical usefulness of the SAEF’s explanations by learners and instructors represent a complementary dimension yet to be empirically explored. The SAEF’s architecture is designed with the explicit objective of supporting learner engagement and fostering transparency; however, these pedagogical benefits are architectural design goals rather than empirically demonstrated outcomes in the present study, which focuses exclusively on computational validation.
The rapid digitalization of health resources highlights the need to understand health information adoption in Indonesia. This study examines trust-related HBM belief pathways and Health Information Access Behavior (HIAB) as both a direct predictor and a moderator. Survey data from 260 respondents across urban, suburban, and rural settings were analyzed using PLS-SEM, with 20 semi-structured interviews used for interpretive context. Perceived benefits, cues to action, and self-efficacy significantly increased adoption, and HIAB had a significant positive direct effect. Moderation was limited: only the perceived barriers and HIAB interaction was significant and negative (β = −0.123, p = 0.023), while the other five interactions were not significant. Thus, access behavior does not uniformly strengthen trust-related HBM pathways; its moderating role is specific to perceived barriers.
Personal health management and clinical health systems are increasingly being transformed by data-driven smart technologies that enable remote monitoring, predictive analytics, and digitally mediated care. Wearable technologies contribute to this transformation by continuously generating wearable health data (WHD), but integrating such data into routine and clinical health management raises governance requirements beyond those for traditional health data. This study examines perceptions of WHD governance among wearable users and health professionals in South Africa. A mixed-methods approach was adopted, with quantitative and qualitative data collected through questionnaires and interviews. The findings show that wearable users are generally confident in WHD governance regarding access, reliability, trust, accuracy, and sharing. Privacy and third-party data practices raised comparatively less concern, though with less certainty, while perceptions of security and the adequacy of current South African laws were less definite and marked more by uncertainty. Health professionals viewed WHD governance more cautiously, emphasising regulatory adequacy, data quality, security, accountability, and the conditions required for meaningful use of WHD in health management. The study concludes that effective WHD governance requires more than formal regulation and must also support intelligibility, credibility, quality assurance, and practical oversight to support the responsible integration of WHD into health management.
Public sector value generation increasingly relies on data provided by information systems, yet the impact of data quality (DQ) on decision-making remains underexamined. While prior research has identified associations between DQ and decision outcomes, strong causal evidence remains scarce. This study addresses the issue in two phases: first, through a scoping review of experimental research on the impact of value-critical DQ dimensions—accuracy, completeness, consistency, and timeliness—and second, through methodological development. Searches across Scopus, Web of Science, IEEE, and PsycInfo, supplemented by citation searching and Google Scholar to avoid inclusion bias, identified 20 studies after screening and full-text review. Rather than excluding studies based on methodological quality, the review critically examined existing experimental designs and their findings. The literature revealed substantial weaknesses, including unclear distinctions between objective and subjective variables, weak manipulations, ambiguous operationalizations, and small sample sizes. Consequently, existing understanding of DQ impact remains limited and often inconclusive. To address these limitations, the article proposes a formal model for DQ experiments and presents a proof of concept for completeness using open data. The findings suggest that completeness influences value appraisal and subsequent choice behavior, particularly under higher uncertainty and larger value differences between alternatives.
Autonomous docking is a core capability enabling full autonomy of unmanned surface vehicles (USVs), whose practical deployment demands visual pose estimation with high efficiency, temporal stability, and closed-loop control compatibility. This paper proposes a lightweight monocular visual docking perception framework based on MobileNetV2 and a temporal convolutional network (TCN). In this framework, a MobileNetV2 backbone is adopted to perform end-to-end regression of the USV’s relative pose with respect to the dock from a single monocular image, while a feature-level TCN module fuses sequential visual features across consecutive frames to enhance the short-term stability of pose estimation. To validate the performance and reliability of the proposed method, a high-fidelity simulation environment is established to conduct closed-loop USV docking tests. Comparative results demonstrate that the MobileNetV2 backbone reduces inference latency compared with the VGG19 architecture, and the embedded TCN module effectively suppresses inter-frame pose fluctuations and abnormal estimation jumps. The proposed method provides an efficient and temporally consistent visual perception solution for simulation-validated USV autonomous docking systems.
The adoption of federated learning (FL) has been on the rise in recent years due to the decentralized approach to data handling. Vertical federated learning is a type of FL that allows different parties to train shared models on complementary feature spaces without the direct exchange of data. However, the gradients these parties exchange can inadvertently carry sensitive information. Adversaries exploit this leakage to mount label inference attacks (LIAs) and adversarial attacks. To curb this, defense mechanisms have been deployed, but most of them either trade robustness for privacy and model utility or vice versa. This study addresses this gap by introducing an improved defense mechanism that combines adversarial training (to harden the model against adversarial perturbations) and differential-privacy-style noise injection (aimed at restoring the label privacy weakened by adversarial training) to collectively enhance the robustness of the existing KDk defense mechanism with marginal model utility trade-off. Instead of relying on heavy encryption or post-processing techniques, it builds privacy directly into the learning dynamics of the model. It was evaluated using five publicly available datasets spanning three data modalities with the proposed mechanism achieving competitive near-baseline accuracy while significantly reducing label-inference success. Under FGSM-based adversarial evaluation, the robustness gap of this mechanism was found to be approximately 1% compared to the 36% robustness gap of the existing KDk mechanism. The Privacy Leakage Index (PLI) reached 81.32%, 96.08%, 82.41%, 86.68% and 73.88% for CIFAR-10, CIFAR-100, CINIC-10, Yahoo! Answers and Criteo datasets, respectively. The results suggest that robustness and privacy security objectives can coexist to secure VFL with minimal effect on model accuracy.
The growing adoption of Augmented Reality (AR) in education raises challenges related to the cognitive load that students experience during learning. Inadequate management of this load can negatively affect information processing during learning activities. Although previous studies have explored AR applications, most have focused on usability and learning outcomes, leaving aside the detailed analysis of Cognitive Load Theory (CLT), particularly its dimensions: intrinsic cognitive load (ICL), extraneous cognitive load (ECL) and germane cognitive load (GCL). To address this gap, this study evaluates cognitive load using the HistARium application, designed to support history learning with interactive experiences. A total of 60 students participated and completed a structured questionnaire to measure ICL, ECL, and GCL after using the application. The results show a positive distribution: moderate intrinsic load, low extraneous load, and high germane load. These findings indicate that the HistARium application balances content complexity, reduces unnecessary effort, and promotes cognitive processes associated with information organization and schema education construction. Furthermore, the results suggest that integrating CLT principles into AR applications supports cognitive processing during learning, while also opening new opportunities for the development of immersive and adaptive learning environments.
Glaucoma is a leading cause of irreversible vision loss, and its early diagnosis remains critically important yet challenging. Traditional assessment based on the cup-to-disc ratio is often insufficient at early stages, whereas the retinal vascular network can provide additional quantitative biomarkers. This study develops and validates a binary classification method for distinguishing healthy from glaucomatous fundus images by combining deep-learning-based vessel segmentation, fractal and multifractal analysis, and textural features. The public ORIGA dataset is utilized. Images are converted to grayscale using three alternative approaches, followed by Gray-Level Co-occurrence Matrix texture analysis and fractal analysis based on the differential box-counting method. Vessel segmentation is implemented via a U-Net neural network trained on a combination of public datasets, after which multifractal analysis is performed on the resulting binary masks. The extracted features are used to train and compare several machine learning models with hyperparameter optimization. The best-performing model among ONH-based features (Random Forest) achieves 75.00%; however, a logistic regression model using multifractal parameters and CDR reaches 86.17%, substantially outperforming the CDR-only baseline (66.15%). Notably, while classical fractal dimension shows only marginal differences (1–2% relative change) between groups, multifractal parameters reveal distinct changes: the multifractal spectrum width Δα increases markedly and the minimum singularity exponent αmin decreases in glaucomatous eyes, indicating increased heterogeneity of the vascular network. These findings suggest that multifractal characteristics of the vascular network can serve as reliable and sensitive biomarkers for automated glaucoma screening, offering clear advantages over classical fractal analysis.
The rapid growth of the green bond market has intensified the need for transparent and reliable monitoring systems, particularly in circular-economy environments characterized by complex, multi-stakeholder, and dynamic interactions. However, existing monitoring approaches still rely heavily on static, issuer-driven disclosures, which sustain information asymmetry and increase the risk of greenwashing. This study systematically reviews the role of digital technologies in enhancing green bond monitoring within circular economy systems. A systematic literature review (SLR) was conducted using the Scopus database, covering publications from 2022 to 2026 and yielding 56 eligible studies. A bibliometric analysis using VOSviewer identified major research trends, thematic clusters, and collaboration patterns within the field. The findings reveal four dominant technological pillars—blockchain, artificial intelligence (AI), Internet of Things (IoT), and digital twin—that support data verification, automated analytics, real-time environmental monitoring, and system-wide integration. Although these technologies show significant potential, the literature remains fragmented and lacks comprehensive monitoring architectures that integrate technological, governance, and regulatory dimensions. This study contributes to the literature by synthesizing these technologies through a business informatics perspective and highlighting digital twin architectures as a promising foundation for integrated green bond monitoring. The findings provide practical insights for regulators, issuers, and investors seeking interoperable, transparent, and trustworthy monitoring ecosystems that strengthen accountability and credibility in sustainable finance.
Chatbots are becoming increasingly significant entry points to digital services and are used in areas including job support, education, healthcare, and customer services. On the other hand, less is known about how chatbots affect people individually, in groups, and in society. Moreover, several obstacles must be overcome before chatbots can realize their full potential. As a result, chatbots have become a significant research topic in recent years. We propose a research agenda outlining future directions and issues to advance knowledge in chatbot research and education. This research involves the quantitative analysis of the impact of these chatbot tools on academic staff, with a count of 40, and students, with a count of 300, at Al Zahra College for Women (ZCW). The links are uploaded electronically in bilingual form for both staff and students, and the responses are retrieved from them. This analysis is conducted in IBM SPSS Statistics version 29, and a comparative report is also prepared based on the questionnaire responses from students and staff members of ZCW. The study investigates the effects of artificial intelligence (AI) chatbots on the faculty and students at ZCW. The use of AI-powered tools in educational settings is examined, along with their effects on administrative, instructional, and learning procedures. This will enable us to identify the advantages of using AI tools within E-Learning systems. The results demonstrate how well AI chatbots can streamline administrative duties, enhance student involvement, and provide academic help. However, there are drawbacks as well, such as user adjustment, privacy issues, and technological constraints. The study offers helpful suggestions for enhancing chatbot integration at Al Zahra College to enhance learning results and operational effectiveness.
Modern inpatient care generates irregular streams of heterogeneous clinical events, yet most predictive models require fixed feature matrices, predefined time windows, or discretization of continuous measurements. We developed CHaRT, a decoder-only autoregressive transformer designed to jointly forecast the identity of the next clinical event and, when applicable, its associated continuous value. CHaRT was trained and internally validated on structured electronic health record data from adult acute-care encounters across a 12-hospital health system in Minnesota from 2001 to 2025. The final corpus included 4,447,625 encounters from 1,301,502 patients and 701,556,877 non-padding clinical event tokens spanning vital signs, laboratory values, medications, diagnoses, microbiology, virology, imaging, fluids, and outcomes (ICU transfer or death). Encounters were split into training, validation, and test sets before vocabulary construction, normalization, and windowing. On the held-out test set, CHaRT achieved Top-1, Top-5, and Top-10 next-event accuracies of 51.61%, 87.34%, and 93.22%, respectively, with perplexity 4.50 and expected calibration error 0.0109. For numeric prediction, z-score MSE was 0.3812 for vital signs and 0.5713 for laboratory values. Seeded examples generated clinically coherent trajectories. Using model representations, a linear probe predicted deterioration (ICU transfer or in-hospital death) at a 6 h landmark with AUROC 0.95–0.97, indicating that learned representations transfer to downstream clinical risk prediction.
Personalised three-dimensional (3D) facial avatars underpin a wide range of immersive virtual, augmented, and mixed reality (XR) experiences, yet conventional 3D capture pipelines remain prohibitively expensive and computationally demanding for prototype-stage XR applications. This study presents and evaluates a lightweight hybrid 3D Morphable Model–Convolutional Neural Network (3DMM-CNN) pipeline that reconstructs an animation-ready 3D facial mesh from a single unconstrained RGB photograph and exposes it through an interactive prototype with native export to XR-ready asset formats. A four-channel ResNet-50 backbone fuses RGB pixels with a landmark-mask channel, regresses the 3DMM shape, expression, pose, and illumination parameters, and is refined through a multi-task loss that combines 3D parameter regression, 2D landmark consistency, and image-to-mesh-to-image cycle consistency. The model is trained on a curated 2000-image subset of the LFW-People corpus and evaluated under four yaw-angle strata. The results indicate that on a held-out 400-image test set, the pipeline attains R2 = 0.854, MSE = 0.022, Pearson r = 0.92, and MAPE = 10.6%, with a single-frame inference latency of 35 ms on a commodity RTX-class GPU. Robustness to head rotation improves by 29.9% at extreme poses (60–90° yaw) compared with a single-modality baseline. A Blender-integrated prototype successfully exports the reconstructed mesh as a deformation-ready asset for Unity- and Unreal-based XR engines. The proposed pipeline offers a cost-effective, real-time-capable component for XR avatar prototyping, lowering the entry barrier for small studios, immersive-learning developers, and AR/MR telepresence research. On the standard AFLW2000-3D benchmark, the pipeline additionally attains a Normalised Mean Error of 2.47% and a full-vertex reconstruction error of 1.50%, which is competitive with published lightweight baselines while retaining sub-50 ms inference latency.
As online luxury fashion resale platforms continue to emerge and expand, persistent challenges related to trust and product authenticity have become increasingly pronounced. digital product passports (DPPs) offer a potential solution to these trust-related challenges by providing detailed information about a product, thereby enhancing product transparency and traceability throughout the product lifecycle. Drawing on signaling theory, this study investigates how DPP-enabled platforms influence consumers’ purchase intentions toward luxury fashion resale consumption. An online survey generated 307 valid responses. Structural equation modeling (SEM) was employed to test the proposed hypotheses. A multi-group chi-square difference test was also conducted to test the moderating role of DPP usage experience. The results demonstrate that trust and attitude are decisive predictors of consumers’ intention to purchase second-hand luxury fashion products from DPP-enabled platforms. Risk reduction alone does not directly drive intention, highlighting the distinction between eliminating uncertainty and fostering positive motivation. This study focuses on consumer perception and intention regarding DPPs within the luxury fashion resale market, an area that has received limited empirical attention. By integrating signaling theory with fashion resale and digital innovation, this study offers novel insights into the role of technological transparency in driving consumer engagement in circular fashion.
Causality extraction is an important task in natural language processing, yet it remains underexplored in informal Arabic social media text, particularly in dialectal contexts. This study investigates causal-reason extraction from Saudi Arabic tweets related to sick-leave requests. A gold-standard dataset was annotated for multiple causality-related tasks, including cause-presence detection, cause-span extraction, cause-category classification, causal-marker detection, and causal marker text identification. The study compares two modeling paradigms: fine-tuned BERT-based models, represented by SaudiBERT and AraBERT, and prompting-based large language models (LLMs), represented by GPT-4.1-mini and Gemini-2.5-flash. The descriptive analysis showed strong class imbalance, substantial implicit causality, and uneven cause-category distributions. Results showed that SaudiBERT generally outperformed AraBERT when macro-level and minority-class performance were considered. Among LLMs, Gemini-2.5-flash achieved the strongest overall performance, particularly under natural 10-shot single-tweet prompting, while balanced few-shot prompting improved macro-F1 for cause-category classification. However, step-wise prompting did not consistently improve performance and may have introduced error propagation. Overall, the findings show that causality extraction in informal Saudi Arabic remains challenging, especially for implicit causal expression. The study highlights the complementary strengths of dialect-specific transformers and LLM-based prompting for Arabic causality extraction.
Background: Although pre-exposure prophylaxis (PrEP) is highly effective for HIV prevention, identifying individuals who may benefit from PrEP and delivering personalized prevention recommendations remain challenging in routine and digital health settings. Objective: This study aimed to develop and preliminarily evaluate an integrated artificial intelligence framework combining machine learning (ML) for HIV risk stratification and generative artificial intelligence (GenAI) for personalized PrEP recommendation support. Methods: A curated dataset of 2000 de-identified client profiles from Love2Test platform was used for proof-of-concept model development. Profiles were labeled as low or high HIV acquisition risk by domain experts based on structured behavioral information. Multiple ML classifiers were trained and compared using PyCaret. The selected model was integrated with a generative AI model through structured prompting to generate personalized PrEP recommendation content. The integrated framework was evaluated through structured physician assessment by four independent medical doctors. Results: The selected model showed strong internal discrimination for classifying high versus low HIV acquisition risk. The integrated framework also received favorable physician evaluation for clinical accuracy, explanation validity, contextual relevance, and error minimization across fixed and randomly selected profiles. However, because expert labeling was based on structured behavioral indicators closely related to the model inputs, the high internal performance should be interpreted within the context of this proof-of-concept study. Conclusions: The proposed framework provides a structured approach to support HIV risk stratification and personalized PrEP recommendations in a clinician-aligned manner. However, this study was an offline proof-of-concept and did not directly evaluate patient interaction, PrEP uptake, stigma, adherence, or clinical outcomes. Prospective studies using larger and more representative real-world datasets are needed to assess implementation, generalizability, and impact on service engagement and PrEP initiation.
Hyperspectral imaging (HSI) provides rich spectral information and serves as a non-destructive technique for forensic stain analysis. Conventional approaches often exhibit degraded performance due to the high dimensionality and spectral redundancy inherent in hyperspectral data. To address this challenge, a hyperspectral dataset comprising nine beverage stains-papaya, coffee, pomegranate, orange, tea, wine, whisky, rum, and brandy-is developed. Building on this dataset, an ensemble framework that combines an optimized autoencoder (AE), channel-attention (CA)-enhanced one-dimensional convolutional neural networks (1D CNNs), and a Limited Memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS-B)-based weighted fusion strategy is proposed. The autoencoder learns compact latent representations from the 204-band hyperspectral vectors, reducing redundancy while preserving discriminative spectral features. CA emphasizes informative spectral bands and improves stain separability. Multiple 1D CNN models are trained using different latent dimensionalities, and their class probability outputs are fused through an optimized L-BFGS-B weighting scheme, where higher-performing models contribute more strongly to the final decision. Experimental results demonstrate classification accuracies of 96.54%, 97.19%, and 97.86% for the AE32 CA, AE64 CA, and AE128 CA models, respectively, with the optimized ensemble achieving an accuracy of 98.28%. Additionally, the time-dependent evolution of beverage stain reflectance is systematically analyzed using overlapped, normalized reflectance signatures acquired at time intervals of 0 min, 1 h, 2 h, 3 h, 4 h, and 5 h. The results confirm that AE-based latent compression, CA, and L-BFGS-B optimized ensemble fusion enhance hyperspectral beverage stain classification, providing an effective and extensible framework for forensic trace evidence analysis.
Bees are essential pollinators for agricultural systems, making accurate, automated monitoring of their behavior critical for assessing colony health and ecosystem stability. Recent advances in computer vision and artificial intelligence have enabled large-scale bee traffic monitoring at hive entrances; however, most existing event classification methods focus exclusively on simple entrance and exit events. This simplification overlooks compound movements—such as U-turns and guarding behaviors—that represent a substantial portion of bee activity and can lead to inaccurate trajectory reconstruction and misleading behavioral interpretations. In this work, we systematically analyze existing event classification strategies used in automatic bee traffic monitoring, evaluating their performance on both simple and compound movements. We then propose extended classification methods that explicitly model compound events by incorporating bidirectional movement patterns derived from positional and angular cues. Using a manually annotated dataset of computer-vision-based hive entrance recordings, we compare threshold-based, displacement-based, and angle-based approaches under simple and mixed-event conditions. Our results demonstrate that compound events account for over one-third of all detected movements and that classification methods explicitly designed to handle bidirectional behavior substantially outperform traditional approaches in both accuracy and robustness. In particular, threshold-based bidirectional classification achieves near-perfect performance when full trajectories are available, while displacement-based methods provide a reliable alternative under partial observations. These findings highlight the importance of modeling compound behaviors in automated bee monitoring systems and contribute to more accurate flight reconstruction, behavioral analysis, and AI-driven decision support for precision agriculture and pollinator management.
