The halal industry is a contemporary global trend of economic development because of the growing number of Muslim consumers. This is realized in Thailand, although it is a Muslim-minority country. Research studies have been conducted to support the readiness of Thailand as a regional hub of the Halal industry. However, a few works emphasize English studies on Thailand’s Halal industry. Thus, this study aims to analyze lexical, grammatical, and semantic features of English regarding the Halal industry in the Bangkok Post, a leading English-language daily newspaper in Thailand. It also discusses the extent of the structural linguistic features that indicate and contribute to World Englishes in the Islamic context of the Thai and global economy. Based on the frameworks of morphology, grammar, semantics, and World Englishes using textual analysis, results revealed that the journalists utilized and constructed a range of outstanding lexical formations as well as phrases and clauses in English that incorporate Arabic and Thai and convey specific semantic features in order to present the Halal industry in Thai society. These structural linguistic features can indicate their Islamic English, World Englishes of global commerce/economy, and Thai Muslim identity of English.

Fractal analysis and magnetic resonance imaging (MRI) semantic features to identify intracranial solitary fibrous tumours and atypical meningiomas.

Research shows that as toddlers' vocabularies expand, words in the early lexicon become increasingly interconnected through shared phonological and semantic features. Understanding how these dimensions jointly shape lexical organization is central to theories of early spoken word recognition. The present study investigated how the simultaneous presence of phonological and semantic similarity between nouns influences lexical activation during spoken word recognition. We presented 21-month-old English monolinguals with an intermodal preferential looking task adapted to a priming paradigm while their eye movements were recorded with an eye-tracker. Participants heard a spoken noun (prime) followed by a related or unrelated spoken noun (target). The experiment included three conditions: Phonologically Related, where prime-target pairs share the initial phonemes (e.g., toe-toast); Phono-Semantically Related, where prime-target pairs share the initial phonemes and belong to the same semantic category (e.g., turkey-turtle); and Unrelated, where prime-target pairs do not share the initial phonemes and do not belong to the same semantic category (e.g., bubble-toast and box-turtle). Results revealed two key findings: (1) Targets in the Phonologically Related condition elicited significantly fewer looks than the Unrelated condition, suggesting phonological interference. (2) Targets in the Phono-Semantically Related condition elicited significantly more looks than both the Unrelated and Phonologically Related conditions, indicating strong facilitation when both cues are present. Additionally, girls demonstrated more pronounced word recognition than boys. This study extends our understanding of the interactive roles of phonological and semantic cues, as well as sex differences, in mental lexical organization among young toddlers.

Harnessing multimodal emotion features in depression detection across gender: Integrating large language model, acoustic fusion and facial expression recognition.

Zero-shot fault diagnosis for gas turbines via alignment of generated semantic samples and data features

Detecting deepfake videos remains a challenging task, especially in scenarios involving unknown manipulation methods or unseen data distributions. Most existing video deepfake detection methods rely on high-level semantic features, which often lead to overfitting of facial identity information and poor transferability. In this work, we explore a novel perspective by modeling videos through 3D differential operations along temporal and spatial dimensions. To exploit the spatial–temporal variation information of the video content, the proposed approach decomposes videos into single-axis 1D differential signals, which are then transformed into 2D representations for efficient learning. This procedure enables the use of lightweight 2D CNNs while retaining directional forgery cues. Our experiments, aimed at analyzing whether these differential signals capture discriminative patterns useful for distinguishing real from fake content, show that the proposed method achieves strong intra-dataset performance and reveals complementary information across dimensions. These findings suggest that differential signals could potentially support generalization when integrated into broader detection frameworks. • We propose 3D Differential Decomposition modeling for deepfake video detection. • Multi-directional and multi-order differential operation are considered. • Optimization for differential order selection and fusion strategy are explored.

• Two publicly available realistic 3D semantic change detection datasets. • A novel multi-task learning architecture for large-scale 3D point cloud semantic change detection. • A multi-dimensional change encoding module that refines the cross-temporal neighborhood variations. • A change-guided semantic refinement module that enhances the representation of semantic features. • A semantic-awareness change interaction module that complements the characteristic of change. 3D semantic change detection enables the detection and identification of both changes in urban objects and their semantic categories, providing fine-grained change information for downstream applications. Existing methods rely on single-branch architectures with predefined output labels, which is simple but suffers from complex output definitions and ineffective multi-task coupling, compounded by scarce annotated 3D realistic data. To overcome these challenges, firstly, two 3D realistic semantic change detection datasets are constructed and published, named HKSCD and UtrechtCD , which utilize oblique photogrammetry point clouds and LiDAR point clouds to describe 9 semantic categories and 2 change types in Hong Kong, China, and Utrecht, Netherlands, covering 15 square kilometers with 370 million points. Secondly, a Multi-task Interaction Siamese Network (MISNet) for 3D point cloud semantic change detection is proposed. It deeply couples semantic segmentation and change detection, enabling the simultaneous prediction of both tasks within a unified architecture. The proposed multi-dimensional change encoding module computes cross-temporal neighborhood relationships from multiple dimensions to extract accurate point cloud change features. Additionally, the change-guided semantic refinement module and the semantic-awareness change interaction module leverage change information to support semantic consistency and utilize semantic information to assist inter-class change detection to promote cross-task consistent modeling underpinned by the cross-learning strategy. Extensive experiments demonstrate that MISNet achieves mIoU of 84.15% (HKSCD), 85.15% (UtrechtCD), and 89.58% (Urb3DCD-V2), outperforming existing methods by + 2.21%, +1.43%, and + 1.46%, respectively. The code and dataset are available at https://github.com/zhanwenxiao/UrbanSCD and https://github.com/zhanwenxiao/MISNet .

Vision-Language Transformers (VLTs) have achieved remarkable success, yet their high computational costs remain challenging due to numerous input tokens and large model parameters. Existing VLT compression methods primarily rely on single-modality-based token pruning or coarse-grained weight pruning techniques. However, these methods face significant obstacles, such as ignoring the critical alignment of different modalities and lacking layer-wise dynamic token pruning flexibility, exhibiting inevitable performance degradation due to coarsegrained weight pruning, and struggling with the simultaneous compression of both input tokens and model parameters. To address those limitations, we propose MADTP++, a novel approach that integrates custom-made token and weight pruning processes into a unified framework, achieving superior compression in both parameter counts and computational costs. Specifically, for the token pruning process, we introduce the Multi-modality Alignment Guidance (MAG) module and the Dynamic Token Pruning (DTP) module to align semantic features across different modalities and guide the dynamic elimination of redundant tokens based on different input instances. For the weight pruning process, we propose a Hardware-aware Weight Pruning (HWP) module that leverages the Sparse Tensor Cores across diverse hardware setups to enable fine-grained parameter pruning within VLTs. To further unify token and weight pruning, we also propose a Cooperative Optimization Training Strategy that automatically allocates GFLOPs and parameter reductions per branch before pruning and employs Knowledge Distillation Constraints to facilitate joint optimization of both pruning dimensions. Extensive experiments conducted on various VLT models and datasets demonstrate that MADTP++ can significantly reduce model parameters and computational costs while maintaining competitive performance.

A Character-level Convolutional Recurrent Interaction Network for joint traditional Chinese medicine clinical named entity recognition and relation extraction.

PointCore: An efficient framework for unsupervised point cloud anomaly detection using joint local-global features.

A DINO-based progressive semantic enhanced infrared and visible image fusion network.

Forensic sketch generation plays a crucial role in criminal investigations where no photographic evidence is available. Traditional sketching methods rely heavily on skilled artists and subjective interpretations of eyewitness descriptions, which often leads to inconsistencies and delays. This project proposes an automated system that leverages Generative AI, specifically diffusion-based models, to generate realistic forensic sketches from textual descriptions. The system utilizes Stable Diffusion XL for high-quality image generation and integrates biometric and semantic feature extraction using InsightFace and BiSeNet. A hybrid matching mechanism using FAISS is employed to compare generated sketches with a mugshot database, providing ranked suspect identification. The proposed framework improves accuracy, scalability, and efficiency by combining text-to-image generation with multimodal face matching, making it a practical solution for modern forensic applications.

ABSTRACT The active expansion of digital media, as well as social networking services, has raised the dispersion of false information and fake news to serious issues, which put pressure on society, its opinion, and democracy itself. Manual verification is not easy and efficient since fake news tend to spread more rapidly than real news because it is sensational. The current research provides NEWSGUARD, a smart system that is aimed at the automated fake news detection relying on a hybrid method of using machine learning and deep learning techniques. The suggested system proposes the use of Natural Language Processing (NLP) in processing of text and extracting features. Term Frequency-Inverse Document Frequency (TF-IDF) is used to create statistical features whereas semantic representations are retrieved with word embeddings. Various classification models are deployed, such as Naive Bayes, Logistic Regression, Support Vector Machine (SVM), Convolutional Neural Networks (CNN) and Long short-term memory (LSTM). The hybrid model is created that combines the benefits of both machine learning and deep learning models and allows achieving better performance in classification. The experimental findings show that the hybrid model has a higher performance compared to the individual models where it has a high accuracy with a high precision, recall and F1 score. This research result demonstrates the usefulness of statistical and semantic features as a combination to detect fake news. The suggested NEWSGUARD solution offers the scalable, correct, and effective tool to fight with the misinformation, and it will be able to be further improved by using advanced models and real-time implementation.

Proverbs are paremiological units that express a people’s worldview, life experience, and moral values in a concise yet profound manner. Among these, proverbs devoted to the concept of “mind” occupy a special place. These proverbs reflect generalized conclusions about human cognitive activity, thinking patterns, behavioral models, and social positions. Proverbs representing the notion of “mind” hold a central position in the paremiological systems of many languages. By expressing collective wisdom briefly and compactly, they encode cultural attitudes toward intellect, reasoning, consciousness, memory, and judgment. This article examines proverbs on the theme of “mind” in terms of their semantic, lexical, syntactic, stylistic, pragmatic, and cognitive aspects. The study demonstrates that proverbs on the “mind” are not only moral and educational tools but also linguistically rich structures that reveal how societies conceptualize human cognition. The article analyzes the lexical, semantic, syntactic, and stylistic features of proverbs about “mind,” exploring their expressive potential and communicative functions.

Ocular diseases have emerged as the leading causes of blindness and low vision, necessitating timely detection and treatment. However, computer-aided approaches face significant challenges in accurately diagnosing these diseases. Specifically, ocular diseases often exhibit a long-tailed distribution, leading to a complex class-imbalanced scenario. Moreover, the coexistence of multiple diseases in a single patient gives rise to a problematic issue of label co-occurrence. In this study, we propose a novel alternate group training strategy as an effective approach to tackle the multi-label long-tailed data distribution problem. Firstly, we partition the long-tailed data into several groups based on semantic feature relations. This division helps reduce the challenges of class imbalance and label co-occurrence. With these groups established, we employ a gradient-based self-weighted loss to train a teacher network in an alternate way. Furthermore, a student model is trained on the original dataset under the guidance of the teacher network, utilizing a weighted class-balanced distillation loss. The class-balanced distillation loss also alleviates the class-wise imbalanced distribution and instance-wise label co-occurrence. Extensive experimental results have demonstrated the superiority of our proposed method which achieves promising performance on the publicly available dataset. In addition, our approach achieves promising performance when expanding the single-teacher model to multiple-teacher models.

Via cross-correlation algorithms or synchronized acquisition of signals, the alignment of heterogeneous data with unknown semantic time shifts and intermittent semantic variations cannot be solved. The shift is caused by different data acquisition principles of sensors, different response discrimination principles using heterogeneous data, etc. Here, we report an unsupervised alignment architecture with a supervised learning model as the kernel to overcome the limitations of brain cognition, perception, and storage in aligning complex heterogeneous data. A set of data with a time shift is input into the kernel model of the architecture to predict the semantic labels, features or continuous values corresponding to another set of data. The time shift corresponding to the maximum testing accuracy or the minimum mean squared error is the alignment parameter for the two heterogeneous datasets. This architecture is expected to serve as a preprocessing step for semantic mining of signals and for information fusion.

Objective To evaluate the diagnostic performance of deep learning−based radiomics (DL) and hand−crafted radiomics (HCR) in differentiating benign from malignant orbital tumors. Methods A retrospective analysis was performed on CT data from 145 patients (48 benign, 97 malignant) diagnosed between December 2014 and March 2024. Two radiologists independently assessed conventional CT semantic features (e.g., lesion location, margin definition, internal density homogeneity, calcification, necrosis, and enhancement pattern). Deep transfer learning (DTL) extracted DL features, while traditional methods were used to obtain HCR features. Feature fusion, selection, and modeling were performed using the Least Absolute Shrinkage and Selection Operator (LASSO). Pathological diagnosis served as the gold standard. Model performance was evaluated using receiver operating characteristic (ROC) curves. A nomogram integrating clinical data and significant semantic features was constructed for visualization. The DeLong test and decision curve analysis (DCA) assessed model effectiveness. Results Multivariate analysis confirmed that homogeneous enhancement and ill−defined/infiltrative margins were independent CT features differentiating benign from malignant tumors. A total of 14 HCR and 30 DL features were extracted; 36 features were retained after fusion. The HCR, DL, fused, and nomogram models achieved AUCs of (0.859/0.816), (0.957/0.826), (0.986/0.811), and (0.975/0.837) in the training and test cohorts, respectively. The DeLong test showed no significant difference between the fused model and the nomogram in either cohort ( P = 0.090 and P = 0.198), whereas differences for other model pairs were significant ( P < 0.05). DCA indicated that the nomogram provided higher clinical utility. Conclusion The fused model outperformed single radiomics approaches in accuracy. The nomogram, which integrates clinical data and semantic features, demonstrated superior predictive performance and may support clinical decision−making, particularly for patients who cannot undergo invasive procedures.

Abstract Speakers use resources to facilitate information processing. Structural markers are information regulators that signal the start and end of information blocks, place them in an order within a sequence, and make it possible to anticipate processing. Their role is especially important in expository textual discourse and in specialized texts where there is a special need to prioritize information to non-specialist audiences. We propose an eye-tracking study of the Spanish structural markers en primer lugar (Eng. ‘firstly’) , en segundo lugar (Eng. ‘secondly’) and por último (Eng. ‘finally’) to verify if their semantic and pragmatic features influence how native speakers process the structures on which they operate. Our starting hypothesis was that their presence may lead to a reduction of the cognitive effort during reading compared to an equivalent text where they are absent. Data analysis from a reading eye-tracking experiment with Spanish native speakers confirmed that the presence of structural markers in enumerative fragments has an accelerating effect on processing during the first reading. However, during the rereading phase this facilitative effect was not found. From this it follows that marking the information structure in an enumeration in specialized texts hardly represents a quantitative advantage for the native reader.

Deep learning is transforming microscopy, yet models often fail when applied to images from new instruments or acquisition settings. Conventional adversarial discriminative domain adaptation (ADDA) retrains entire networks, often disrupting learned semantic representations. Here, we overturn this paradigm by showing that adapting only the earliest convolutional layers, while freezing deeper layers, yields reliable transfer. Building on this principle, we introduce subnetwork image translation ADDA with automatic depth selection (SIT‐ADDA‐Auto), a self‐configuring framework that integrates shallow‐layer adversarial alignment with predictive uncertainty to automatically select adaptation depth without target labels. We demonstrate robustness via multimetric evaluation, blinded expert assessment, and uncertainty‐depth ablations. Across exposure and illumination shifts, cross‐instrument transfer, and multiple stains, SIT‐ADDA improves prediction fidelity and downstream segmentation over full‐encoder adaptation and nonadversarial baselines, with reduced drift of semantic features. Our results provide a design rule for label‐free adaptation in microscopy and a recipe for field settings; the code is publicly available.

Early detection of pulmonary nodules plays a critical role in improving the survival rate of patients with lung cancer. However, existing detection methods often exhibit limited feature stability and inadequate contextual awareness, particularly when handling small nodules or nodules with complex structures. To address these limitations, we propose a 3D detection method, termed TCMNet. The proposed model consists of two key components. The SD module incorporates a dynamic tanh nonlinear transformation into the Swin Transformer to improve semantic retention and feature stability in regions containing small targets. The CMCA module combines cascaded multiplicative connections with a Monte Carlo based attention mechanism to enhance multi-scale feature interaction and increase the focus of the model on pulmonary nodules. Experiments on the LUNA16 dataset show that TCMNet achieves an average FROC score of 91.59%, outperforming several state-of-the-art methods. These results indicate that the proposed architecture provides a robust and effective solution for three-dimensional pulmonary nodule detection.