Search Results (318)

Internal and civil wars affect the lives of religious and ethnic minorities the most. For Syrian citizens of Armenian origin, the Republic of Armenia represented one of the most accessible and meaningful destinations to relocate to, shaped by shared ethnicity, collective memory, and historical ties. When the Syrian war erupted in 2011, thousands opted to resettle in Armenia, yet they and host institutions struggled to categorize them as immigrants, refugees, or repatriates. This ambiguous status has received little scholarly attention. To explore these complexities, the study employed a survey-based research design involving 124 participants, supplemented by an open-ended question intended to capture personal narratives and nuanced identity negotiations. The manuscript examines how the labels immigrant, refugee, and repatriate carry distinct legal, social, and emotional implications, especially against the backdrop of the 1915 Armenian Genocide’s enduring memory and the particularly negative connotations of “immigrant” and “refugee” in Western Armenian and Arabic languages. Within this contested semantic and policy terrain, repatriation appears not merely as a bureaucratic category but as a culturally resonant and sometimes preferred pathway for some Diaspora Armenians, informed by lifelong exposure to repatriation narratives through formal education (language textbooks) and informal communal practices. The case sheds light on the broader conception of stakeholders, including how they self-identify, how they understand their status in Armenia, and the factors shaping their choices, particularly in the context of contemporary geopolitics and the role of education in influencing external perceptions of them. Full article

Trajectory data is valuable for applications such as urban planning, but its public availability is often limited by privacy concerns and data collection costs. While recent diffusion models have shown promise in generative tasks, existing methods rarely integrate personalized conditioning with road network constraints. As a result, they struggle to simultaneously achieve personalized mobility modeling and high road-network spatial validity, resulting in limited trajectory quality. In this paper, we propose Sem-RoadDiff, a symmetry-aware dual-guided diffusion model for personalized and road network-constrained trajectory generation. Specifically, our model incorporates two key components. First, we design a semantic preference guidance mechanism to encode user history into a preference-weighted user embedding using a temperature-scaled softmax. This enables the model to capture user-level mobility patterns without directly using raw trip-level records as generation conditions. Second, we introduce a road-aware mechanism to improve overall spatial validity, employing a spatial validity loss derived from the User Mobility Transition Graph. From a symmetry perspective, Sem-RoadDiff aims to preserve distributional symmetry between real and generated trajectories while respecting the inherent asymmetry of directed road-network transitions. Extensive experiments on the Geolife and Porto datasets demonstrate that our approach improves trajectory distributional fidelity compared with the evaluated baselines and improves road-segment connectivity over the diffusion-based baseline. Full article

Multimedia recommenders can use behavioral records together with visual and textual item information, but unreliable interactions and sparse histories still make user preference modeling difficult. Most graph-based methods propagate messages over observed user–item edges as if all interactions were equally informative, so incidental or semantically inconsistent behaviors may distort the learned representations. The standard recommendation loss also provides limited context for modeling dependencies within a user’s historical sequence. We propose MGDSL, a MGDSL applies a multimodal-aware topology denoising module to calculate edge reliability weights for historical interactions from collaborative, textual, and visual evidence, and uses these weights for reliability-aware historical aggregation. In parallel, a masked self-supervised auxiliary task reconstructs masked items from sequence context, adding supervision for latent preference learning. Experiments on three benchmark datasets show that MGDSL consistently improves recommendation accuracy over competitive baselines, with particularly clear gains on the sparsest dataset. Full article

Grounded in the Cognitive Theory of Multimedia Learning and Cognitive Load Theory, this study examined how cue-based instructional support relates to L2 Chinese vocabulary processing and immediate learning outcomes. Forty intermediate-to-advanced learners studied 24 disyllabic pseudowords under four within-subject conditions: no cueing, verbal cueing (linguistic–semantic support via definitions and collocations), physical cueing (typographical enhancement via bolded targets and underlined contextual words), and full cueing. Eye movements, immediate post-tests, and questionnaires were analyzed. The results revealed selective, measure-dependent effects rather than uniform facilitation. In the Orthographic Choice Task, no cueing outperformed full cueing. In the Semantic Priming Decision Task, verbal cueing yielded a higher accuracy than physical cueing, indicating that linguistic–semantic support benefited initial meaning-related processing more than typographical enhancement. No differences emerged in the Sentence Acceptability Judgment Task. Eye-tracking showed shorter first fixations under physical than verbal cueing, suggesting the limited facilitation of early visual orienting. Full cueing showed no consistent advantage over verbal cueing but elicited larger pupil sizes and longer total fixation durations on targets, indicating additional coordination demands. Learners most often preferred full-cueing materials, yet rated verbal cueing as most helpful. An effective cue-based design should align the cue format and content with the target learning dimension while avoiding unnecessary processing demands. The findings reflect immediate learning under controlled conditions rather than long-term acquisition. Full article

Specialized classrooms in architecture schools serve as the core spatial infrastructure of architectural education, supporting theoretical instruction, design tutorials, model making, and critique sessions. However, existing studios often follow rigid layouts that respond inadequately to the diversity of learning stages and usage scenarios. Based on embodied cognition theory, this study integrates spatial morphological analysis, stated preference surveys, and semantic differential questionnaires to establish a multidimensional evaluation framework covering psychological perception, spatial environment, audiovisual perception, professional facilities, and professional atmosphere. Four spatial types, classified through global case review, were examined under class-time and self-study scenarios. Statistical analyses reveal a contextual fit effect whereby the same spatial type exhibits divergent perceptual performance depending on the usage scenario, and a stage-dependent gradient in which environmental concern shifts from sensory qualities toward functional efficiency as academic level advances. Gender, visit frequency, and duration of stay showed limited explanatory power relative to spatial and scenario effects. Six evidence-based optimization principles and three tiered conceptual schemes are proposed for lower-year, middle- to upper-year, and advanced users, respectively. This study offers an empirically grounded reference for the evaluation and tiered optimization of architecture design studio environments. Full article

The rapid evolution of online learning environments has generated diverse and complex data ecosystems. Recommender systems play a central role in leveraging such heterogeneous data to support personalised learning experiences. However, many deep learning-based recommender systems still rely on identifier-based representations that capture co-occurrence and collaborative patterns while overlooking the semantic information embedded in educational activities and the temporal dynamics of learner behaviour. To address these limitations, this study proposes a collaborative-enhanced semantic–sequential recommendation framework for educational platforms that combines structured semantic representation learning, sequential behavioural modelling, and collaborative preference modelling. The proposed architecture integrates a parameter-efficient MiniLM adaptation strategy to extract semantic representations from structured item-related educational metadata and a bidirectional recurrent encoder to model temporal learning patterns from behavioural logs. A gated fusion mechanism is then used to combine semantic and contextual information into learner representations, which are further integrated with collaborative user–item embeddings for top-K recommendation using a Bayesian personalised ranking objective. Experiments conducted on the EdNet-KT1 dataset under chronological splitting, full-corpus ranking, and fixed candidate-sampling protocols show that the collaborative-enhanced model achieves the highest-ranking performance among popularity-based, matrix factorisation, neural collaborative filtering, recurrent sequential, self-attention sequential, and ablation baselines. The model obtains an NDCG@10 of 0.1344 under full-corpus ranking and 0.5383 under candidate sampling, with statistically significant but practically modest improvements over the strongest baselines. Additional ablation, efficiency, and gate analyses indicate that semantic–contextual modelling is most effective when used as a residual enhancement to collaborative recommendation rather than as a standalone replacement. These results suggest that parameter-efficient semantic–sequential modelling, when combined with collaborative preference signals, offers a promising direction for scalable and evidence-based educational recommender systems. Full article

Closed-set heterogeneous domain adaptation (HDA) for Internet of Things (IoT) intrusion detection aims to transfer detection capabilities across environments that differ in devices, telemetry, feature schemas, attack implementations, label taxonomies, and target supervision availability. Although recent HDA methods report strong performance, their deployment meaning is often unclear because improvements over a weak source-only baseline do not show how much target supervision headroom has been recovered or whether adaptation is preferable to direct target-side labelling under the same budget. This paper presents a controlled, anchor-based benchmark for closed-set HDA in IoT intrusion detection. Edge-IIoTset is used as the main fixed target dataset, with transfer from CICIDS2017, UNSW-NB15, CICIDS2017 + UNSW-NB15, and CICIDS2017 + NSL-KDD under single-source and multi-source settings. The benchmark defines fixed resolved contexts, Intersection and Union representation contracts, a five-class closed-set label contract, leakage-safe preprocessing, and an anchor ladder consisting of source-only, correlation alignment (CORAL), matched-budget target-only, and oracle target-only references. Geometric Graph Alignment (GGA) and the Joint Semantic Transfer Network (JSTN) are evaluated as the primary selected native single-source semi-supervised HDA (SS-HDA) and multi-source semi-supervised HDA (MS-HDA) exemplars, while the Prototype-Matching Graph Network (PMGN) and Conditional Weighting Adversarial Network (CWAN) provide $1:10$ method coverage checks. Each method–context–ratio configuration is evaluated across twenty fixed seeds, and DA-versus-target-only differences are tested using paired seed-level statistical evidence. A compact second-target confirmatory experiment using ToN-IoT assesses whether the qualitative headroom recovery and same-budget deployment patterns remain visible under a different IoT/IIoT target. The results show that primary native HDA can recover substantial source-only-to-oracle headroom, but not uniformly. At the $1:10$ labelled target ratio, GGA recovers $0.633$–$0.835$ of the available headroom across C1–C4, while JSTN recovers $0.776$–$0.897$ in the contemporary-source MS-HDA family and $0.872$–$0.926$ in the mixed-vintage family. Same-budget comparisons show that DA is deployment-competitive only in some contexts; in others, direct target-side supervised learning is stronger. The benchmark therefore shows that closed-set HDA should be evaluated as target-conditioned, context-resolved evidence rather than as a pooled method leaderboard. Full article

Wearables design research often evaluates aesthetic and ergonomic features without capturing their emotional and cognitive effects on user experience and buying decisions. This paper investigates both dimensions for smartwatches as screen-based wrist-worn wearable devices (SBWWDs) among Millennials and Generation Z using Kansei Engineering to structure SBWWD design features into users’ emotional perception and affective preferences. The study examines four hypotheses: (H1a) aesthetic perception differs between Millennials and Generation Z, (H1b) aesthetic perception differs across genders within the same generation, (H2a) ergonomic perception and visual needs for smartwatches’ screen interfaces differ between Millennials and Generation Z, and (H2b) ergonomic preferences differ across genders within the same generation. The research adopts a two-phase design methodology. Phase I-A identifies key aesthetic attributes from market-leading smartwatches and develops controlled design stimuli using AI-assisted concept generation. A questionnaire-based survey captures demographic-linked aesthetic preferences and emotional responses, with emphasis on case shape, strap material, and wearable color, to psychological perception and preference in smartwatch product designs. Phase I-B examines ergonomic interface display preferences relevant to smartwatch screens, including contrast and polarity, using Likert scales and bipolar Semantic Differential Scales. Subsequently, participants evaluate the combined interface features’ stimuli through measures of task accuracy and completion, best/worst interface display selections, eye-tracking metrices analysis, as well as emotional and cognitive arousal provoked by psychological intention using the Self-Assessment Manikin. Further, a full factorial design experiment evaluates the effects of participants’ demographic variables, including generation and gender, as well as smartwatch design features, on aesthetics and ergonomics design perception and preference. Phase II applies Kansei Engineering principles by mapping design features to affective responses of Phase I. Findings provide a structured mapping of smartwatch design perception and preferences across generational and gender groups within the Egyptian market, supporting affective principles in SBWWD design guidelines. The study contributes an evidence-based framework that integrates aesthetic and ergonomic features through Kansei Engineering, aiming to enhance online purchasing in smartwatch devices. Full article

Modern web and tourist maps use pictorial symbols to help users quickly and easily identify Points of Interest (POIs). Pictorial symbols are sometimes misinterpreted due to poor design choices. As a result, it is important to evaluate pictorial symbols with map users. This paper uses an online questionnaire to examine how different graphic parameters—such as frame outline, frame background, frame shape, color hue, and pictogram category (semantic, visual, or arbitrary)—are perceived by map users. The evaluation of pictograms includes three aspects: understanding, to capture the map reader’s opinion; preference, to investigate the map maker’s choice; and appropriateness, to document the evaluation of an existing map. Seven popular Points of Interest (POIs) were selected for the evaluation of pictorial symbols: Hotel, Restaurant, Parking, Museum, Airport, Hospital, and Church. Based on the questionnaire results and the statistical analysis of 520 responses, several conclusions were drawn. Users prefer symbols with a frame outline and a frame background. They also prefer symbols with a white background, which increases contrast and improves legibility. In contrast, users do not have a strong preference for a specific frame shape. In general, users can recognize symbol groups based on frame shape, but the effect is stronger when the color hue appears in the frame background or outline. The statistical analysis demonstrates that perceived appropriateness constitutes an objective measure related to comprehension. Furthermore, appropriateness is independent of the pictogram classification as semantic, visual, or arbitrary. Instead, it is determined by the graphic ability of the pictogram to represent a specific POI. This conclusion reaffirms the importance of designing successful semantic and visual pictograms or adopting those already familiar to map users, as familiarity has also been identified as an important factor by this research. Overall, this paper, based on user evaluations, provides practical insights to improve pictorial symbols on a tourist map. Full article

Autonomous driving planning requires not only accurate trajectory prediction but also coherent semantic alignment across perception, decision making, and motion generation. Existing vision-language-based approaches predominantly focus on improving trajectory accuracy, which may lead to limited behavioral consistency. In this paper, we reformulate planning as a structured autoregressive generation task, where reasoning, actions, and future trajectories are jointly produced from multimodal observations. Based on this formulation, we propose Trajectory-Decision Joint Preference Optimization (TDJPO), which is a rollout-based alignment framework equipped with a unified reward that simultaneously captures physical trajectory quality and decision-level coherence. Starting from a supervised fine-tuned model, we construct preference pairs through stochastic rollouts and optimize the model using direct preference optimization. Experimental results on the NuScenes-TP benchmark demonstrate that our approach consistently enhances both trajectory accuracy and semantic consistency compared with supervised fine tuning, trajectory-only optimization, and lightweight vision-language baselines. These findings emphasize the necessity of jointly aligning physical feasibility and decision-level reasoning for achieving coherent and human-like autonomous driving behavior. Full article

We propose QwenMoE-SC, a semantic communication framework that integrates a Mixture-of-Experts (MoE) Large Language Model with three complementary modules: (1) a Graph Neural Network (GNN)-based Unequal Error Protection (UEP) plug-in that assigns semantic importance scores via syntactic dependency graph message passing for adaptive bit allocation, without modifying the pre-trained LLM; (2) NEFTune noise injection during fine-tuning for channel robustness; and (3) a Communication-aware Direct Preference Optimization (C-DPO) strategy that favors semantically faithful yet token-efficient transmissions. Comprehensive ablation studies on AWGN and Rayleigh fading channels show that each component contributes distinct gains, and their combination consistently outperforms traditional separation-based methods and neural baselines in sentence similarity, BLEU score, and semantic-level BER, with the largest improvements at low-to-mid SNR regimes. QwenMoE-SC can also serve as a semantic interface layer within expert and decision-support systems, enabling robust, bandwidth-efficient communication between data sources, inference engines, and human users. Full article

Background: Mechanistic reporting from large-scale biomedical knowledge graphs remains challenging, particularly when integrating structured graph evidence with large language model (LLM)–based explanation in a reproducible and auditable manner. Existing approaches either rely on manual synthesis of graph-derived results or generate unconstrained narratives that lack traceability to underlying evidence. Methods: We present SemNet Explorer, an evidence-grounded knowledge graph–LLM unified framework for automated mechanistic reporting across biomedical domains using SemNet 2.0, a PubMed-scale heterogeneous knowledge graph. Given a set of target concepts and a selected semantic layer, the framework organizes graph-derived evidence into structured regions and generates two complementary report types: global reports for process-level mechanisms and anchor-centric reports for localized mediator-based explanations. A central methodological contribution is an ablation-derived adaptive grounding policy: we systematically compare alternative evidence-integration strategies across report types, semantic layers, and region structures, and use the resulting preferences to guide prompt selection in the deployed system. Results: SemNet Explorer produces stable region decompositions and interpretable report scaffolds across molecular (AAPP), disease-level (DSYN), and pharmacologic (PHSU) representations. For global reports, explicit evidence grounding improves expression quality more consistently than content accuracy, with benefits dependent on evidence density and semantic abstraction. In contrast, anchor-centric reports show consistent improvements in both content and expression under stronger, mediator-constrained prompting. These findings are supported by both pairwise ablation comparisons and absolute score analyses. Conclusions: SemNet Explorer establishes a generalizable unified framework and interactive platform for transforming knowledge graph evidence into reproducible mechanistic narratives across biomedical domains, including multimorbidity analysis, comparative pathophysiology, drug repurposing, and adverse event discovery. The results demonstrate that effective knowledge graph–LLM integration requires adaptive, context-dependent evidence grounding rather than fixed prompting strategies. Full article

End-to-end autonomous driving remains fragile in long-tail scenarios, while incorporating vision-language models (VLMs) introduces substantial deliberation latency that cannot interfere with the real-time planning loop. We present ThinkDrive, a dual-process driving framework designed under explicit real-time queuing constraints. The framework contains four coordinated components. First, a Scene Complexity Estimator regulates System-2 activation through a trigger cool-down mechanism, allowing at most one asynchronous request every $\lceil L_2/\Delta t\rceil$ frames and thereby preventing queue saturation under a System-2 latency of $L_2=565$ ms. Second, a multi-modal System-1 planner generates $K_1=5$ candidate trajectories within 44 ms and is trained with winner-takes-all imitation learning together with explicit score supervision. Third, a two-stage Causal-CoT module uses the VLM to identify risk agents and predict a preferred spatial goal $G_{\mathrm{VLM}}$, after which a single batched scm_rollout selects the safest candidate and extracts its endpoint as a world-coordinate goal anchor ${\mathbf{g}}_{S2}$. Fourth, a Goal-Anchor Replanning module transforms ${\mathbf{g}}_{S2}$ into the current ego frame and selects the candidate whose waypoint at the remaining time horizon is closest to the transformed goal. This design avoids coordinate-space mixing, mitigates bias caused by mismatched temporal horizons, and prevents semantic instability across replanning cycles. On nuPlan test14-hard, ThinkDrive with InternVL2-8B and a 6.8% trigger rate achieves 74.9 PDMs, outperforming AdaThinkDrive at 73.1 while maintaining a nominal latency of 44 ms. Full article

Traditional urban landscape evaluations have primarily relied on either objective spatial metrics, such as the Green View Index (GVI), or subjective human surveys, often failing to capture the complex mechanisms of human environmental perception. This study proposes a novel Explainable Artificial Intelligence (XAI) framework that integrates objective physical configuration with subjective cognitive assessment to predict human landscape preference. Utilizing 159 urban landscape images, we extracted physical features via semantic segmentation (SegFormer) and psychological perceptions via a zero-shot vision-language model (CLIP). Our hybrid Random Forest model successfully bridged these dimensions, achieving moderate yet promising predictive performance (Rsquare = 0.442). SHAP (Shapley Additive exPlanations) analysis revealed that psychological perceptions—specifically Safety (0.104), Fascination (0.096), and Tranquility (0.080)—outperformed traditional objective metrics like GVI (0.067) in determining overall preference, while sub-model interpretation linked these psychological responses to specific physical elements such as buildings, sky openness, low vegetation, and water bodies. The findings suggest that urban green space design should move beyond maximizing greenery quantity and instead prioritize spatial compositions that induce psychological security, visual interest, and restoration. The proposed framework offers a scalable and interpretable tool for human-centered landscape assessment, while acknowledging limitations related to sample size, cultural generalizability, pretrained model bias, and reliance on static two-dimensional imagery. Full article

Autonomous ground vehicle navigation in dynamic real-world environments demands path planning systems that simultaneously accommodate real-time environmental hazards and diverse user-defined objectives requirements that classical algorithms, with their static, single-objective cost functions, cannot fulfil. This paper presents the Semantic Personalised Path Planning (SPPP) system, centred on a novel Semantic Personalised Cost (SPC) algorithm that augments the A* search framework with a dynamically computed personalised cost term. The SPC function integrates eight real-time semantic obstacle categories including traffic congestion, weather severity, road surface conditions, and construction activity with eight user-defined preference dimensions spanning safety, travel time, emergency response, comfort, and battery efficiency. An adaptive scaling mechanism amplifies obstacle penalties near the goal, and a gradient-based weight evolution rule refines preference weights iteratively over successive route segments. The user-defined preference activation directly personalises the routing objective to individual operational needs, with the gradient-based evolution further refining preference alignment over successive route segments. Experiments were conducted in two phases: 500 randomised obstacle configurations on a controlled $8\times 8$ grid, and a real 847-node road graph extracted from OpenStreetMap around SRM Institute of Science and Technology, Kattankulathur, representing a single 1.4 km urban corridor, with obstacle scores derived from live Mapbox Traffic and OpenWeatherMap application programming interface data. Under the full emergency preference scenario, SPPP achieves 94.3% obstacle avoidance versus 31.7% for the Euclidean distance threshold A* baseline, a difference statistically significant at $p < 0.001$ under the Wilcoxon signed-rank test with Cohen’s d ≈ 18.9. Real-world computation time of 1.91 ms on a standard laptop and 3.76 ms on a Raspberry Pi 4 confirms deployability on embedded autonomous vehicle hardware. Full article