ABSTRACT The geological reports and maps accumulated during geological surveying and mapping harbor rich expert knowledge and metallogenic clues. However, efficiently integrating and mining structured knowledge from complex multimodal data of polymetallic deposits remains a critical bottleneck in intelligent mineral prediction. To address this, we propose a knowledge graph (KG)‐enhanced multimodal retrieval‐augmented generation (RAG) framework, Geo‐MAG, for geological map understanding. Specifically, the framework first processes textual geological reports and constructs a structured KG. Concurrently, a vision large model parses geological maps to extract metadata, including legends, geological structures, strata, and lithologies. Leveraging this metadata, relevant subgraphs are retrieved from the KG to facilitate text–map semantic alignment and enhance background geological knowledge. Finally, the integrated map information and structured subgraphs of KG are fed into the GPT‐4o to enable deep semantic interpretation. Experimental results demonstrate that integrating the knowledge graph significantly boosts the GPT‐4o's reasoning capability and interpretability in geological map understanding. The model achieves 77.2% accuracy in geological reasoning tasks, outperforming the direct end‐to‐end GPT‐4o interpretation by 53.7% and lightweight schemes on the basis of basic metadata by 37.4%. This work represents a pioneering application of KG and RAG in geological map understanding, highlighting the synergistic advantages of integrating text and maps, and offering a novel perspective on multimodal integration within the geoscience domain.

Recently, publishers of offensive comments are increasingly employing strategies such as metaphors, abbreviations, and homophones to obscure the aggressive nature of their comments. These strategies pose a significant challenge for existing detection models. At present, many studies mainly focused on the detection of explicit offensive speech, and there were few studies on implicit offensive speech. Our research aims to analyze implicit offensive speech on Chinese social platforms and achieve high detection performance. Firstly, we have collected data from one of the largest Chinese social networking platforms, Weibo, and constructed the first Chinese implicit offensive speech dataset, which contains 54,714 comments. Subsequently, we introduce Enhanced-BERT-Mate-Ambiguity (EBMA), a novel fuzzy semantic interpretation framework that leverages BERT and knowledge graphs. Specifically, this model detects implicit offensive speech by extracting semantic, emotional, metaphorical, and ambiguity features. Finally, extensive experiments were conducted, including comparison tests, robustness tests, and ablation studies, to validate our approach. We tested our model against state-of-the-art models in the field, and an accuracy of 95.83% and an F1-score of 95.52% confirmed its best performance. The performance of our model is visually illustrated through visualization. Moreover, we provide an analysis of error cases to explore the limitations of our model.

This paper examines the properties of algebraically stabilized linear transformations in artificial neural networks (ANNs) under a change of basis and their compatibility with symmetry-invariant architectures. We investigate the conditions under which a stabilizing transformation of the form W^k=I+μD, defined over integer matrices, preserves its properties under conjugation, i.e., when passing to an equivalent form W'=B^(-1) WB in a new basis. It is shown that such transformations form a closed class with respect to conjugational equivalence and can be employed in environments with diverse geometric and symmetry requirements. Additionally, we analyze the role of symmetric actions on the input space and define conditions for the invariance of transformations with respect to specific groups of transformations. Both formal algebraic properties and the geometric and semantic interpretation of stability in this context are considered. Different types of stability are presented – algebraic, conjugational, and symmetry-based – along with the relationships between them. The results are relevant to the design of ANN architectures with predictable behavior, robustness under transformations, and compatibility with predefined symmetric structures.

[18F] fluorodeoxyglucose positron emission tomography - computed tomography (FDG PET-CT) is increasingly used for staging of breast cancer in the primary and recurrent setting, as well as in evaluating treatment response and in follow-up. Quantitative parameters derived from the primary tumor, even in non-metastatic patients (i.e., without distant metastases but possibly with nodal involvement), have shown prognostic value. Beyond visual interpretation, quantitative evaluations may improve diagnostic accuracy and reproducibility. However, current studies often rely on predefined parameters such as maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV), and total lesion glycolysis (TLG), which may overlook the high-dimensional patterns inherent in FDG PET-CT. To address this, we conducted a deep-learning-based analysis of FDG PET-CT from a large retrospective cohort of non-metastatic breast cancer patients, evaluating prognostic value from multiple perspectives. To improve patient prognosis and risk stratification, we developed a multi-omics prognostic stratification (MOPS) model that integrates clinical data, FDG PET-CT, and corresponding medical reports using CMA and transformer-based architectures to predict overall survival (OS) and disease-free survival (DFS). To support clinical applicability, we incorporated interpretability into the model, providing causal explanations, visualization-based insights, and semantic interpretations to help clinicians understand and apply the predictions transparently. The MOPS model markedly improves survival prediction, outperforming single-omics models, TN staging, and molecular subtyping, with C-index values of 0.75 (95% CI: 0.69-0.81) for OS and 0.71 (95% CI: 0.65-0.77) for DFS.

Accurate flood extent mapping plays a pivotal role in water resource assessment, hydrological modeling, and emergency response. This study develops Edge-Guided Attention Fusion Flood Net (EAFFNet), a deep learning model for precise flood extent extraction from satellite imagery. EAFFNet combines spatial feature extraction with semantic interpretation - its spatial branch captures directional features while the semantic branch models long-range pixel relationships. The integrated edge enhancement and attention fusion modules effectively address semantic overlap issues, enabling accurate segmentation of complex flood boundaries. The comparative results on two open-sourced datasets, the Sen1Floods11 and GF-FloodNet datasets, indicated that the proposed model achieved mean Intersection over Union (mIoU) scores of 41.45% and 85.05%, respectively, which perform better than other advanced deep learning models. As a whole, we believe the proposed model can perform well in flood monitoring and provide a reliable technical solution for disaster early-warning systems.

Introduction. In the present-day scientific discourse, there is a great number of theoretical research focusing on the problem of objectifying the semiotic nature of law and analysing the functional construct of legal semantics. Whereas, many practical legal issues, such as: interpretative ambiguity in the meaning-formation and meaning-application of normative acts, lexical vagueness and contextual dependence of legal notions and the incoherence of legal terminology across different legal systems, remain neglected, which leads to contradictions and inaccuracies in legal practice. The aim of the study is to define the methodological principles fostering establishment of the acceptable scope of semantic interpretation of legal notions in the context of building a legal thinking culture. Materials and Methods. The research methodology was based on the principle of jurisprudential definition of legal norm meaning-formation in socio-legal discourse. Analytical, systematizing and pragmatic methods were used to reveal a complex nature of the semantics of law in the context of legal thinking development. The semiotic analysis of the objectivity and normativity of legal notions taking into account the contextual differences of legal definitions, was used as a specialised research method. Results. It was established that normative notions are the complex semantic constructs encompassing a conceptual sphere (normativity) and social reality. For building sustainable models of legal behaviour and legal culture, it is necessary to overcome external and internal conflicts in interpretation of law. In this regard, a number of advisory measures were proposed aimed at establishing acceptable scope of semantic interpretation: differentiation between the informational nature of prescriptive and descriptive notions, semantic monitoring of legal phenomena, and implementation of the principle of discourse contextualism, which makes it possible to formulate the normativity of law requirements based on the specific contextual interpretations. Discussion and Conclusion. A justified conclusion about possibility of a properly selected semantic toolkit to determine the objectivity of perception of the legal norms and, consequently, to improve the process of building a legal culture was drawn. The main advantage of the principle of discourse contextualism such as conjunction of the semantics and pragmatics of legal notions was identified, which provides a fruitful foundation for further theorizing on the nature and metaphysics of law.

Figure-ground reversal (FGR) transcends visual conventions by reversing the roles of figure and ground in brand logo designs. In this research, the authors study how FGR logos affect consumers’ brand attitudes. Using traditional self-reported measures as well as biometric technology, they illuminate the unique nature of FGR’s underlying mechanism and identify moderators to shed additional light on that process. Specifically, they find that the positive effect of FGR logos on brand attitude is mediated by engagement and aesthetic appeal, and moderated by the visual identification and semantic interpretability of FGR objects. Across a multi-method investigation that includes live bidding, incentive-compatible willingness-to-pay, eye-tracking, and multiple boundary condition experiments, the authors provide empirical support for these effects and reveal the underlying mechanism. They conclude by discussing the contributions of the research to the literature on visual marketing phenomena and the implications of the findings for better visual branding in the marketplace.

Internet buzzwords are an important reflection of social mood and cultural mentality. This study employs content analysis and inductive reasoning, complemented by quantitative trend analysis and qualitative semantic interpretation. Using public microblog data, it investigates the transmission trajectories and semantic evolution of two representative internet buzzwords, "yyds" and "lie flat," through trend charts, word clouds, and thematic content analysis. Data collation found that "yyds" presented a pulse communication mode, relying on external hot events to promote, "lie flat" accompanied by the evolution of social issues, and the heat was highly synchronized with public policy discussions. The final research conclusion points out that there are essential differences between the two in communication power, semantic function and mentality mapping: "yyds" reflects the emotional identity and follow-up under consumerism, and "lie flat" reflects the alienation and resistance mentality of young people under social pressure, and jointly reveals the multiple roles of network catchwords as a social mood refractor.

The article examines the duration of the muscle contraction underlying the eyebrow lift, which varies depending on the specifics of the situation: at the beginning of a communicative episode, this movement turns out to be longer than in the process of subsequent interaction. The study of the joint manifestation of various facial expressions revealed the presence of universal patterns inherent in all three analyzed cultures. The most typical facial component accompanying the raising of eyebrows is a smile. The antipode of the so—called “eyebrow flare”, both from the point of view of neuromuscular organization and from the point of view of semantic interpretation, is the work of the muscle, which ensures lowering of the eyebrows and their displacement to each other.

Research on metaphor has steadily increased over the last decades, as this phenomenon opens a window into a range of linguistic and cognitive processes. At the same time, the demand for rigorously constructed and extensively normed experimental materials increased as well. Here, we present the Figurative Archive, an open database of 996 metaphors in Italian enriched with ratings and corpus-based measures (from familiarity to semantic distance and preferred interpretations), derived by collecting stimuli used across 11 studies. It includes both everyday and literary metaphors, varying in structure and semantic domains, and is validated based on correlations between familiarity and other measures. The Archive has several aspects of novelty: it is increased in size compared to previous resources; it offers a measure of metaphor inclusiveness, to comply with recommendations for non-discriminatory language use; it is displayed in a web-based interface, with features for a customized consultation. We provide guidelines for using the Archive to source materials for studies investigating metaphor processing and the relationships between metaphor features in humans and computational models.

Accurately delineating soil elements from satellite imagery is fundamental for regional geological mapping and survey. However, vegetation cover and complex geomorphological conditions often obscure diagnostic surface information, weakening the visibility of key geological features. Additionally, long-term tectonic deformation and weathering processes reshape the spatial organization of soil elements, resulting in substantial within-class variability, inter-class spectral overlap, and fragmented structural patterns—all of which hinder reliable segmentation performance for conventional deep learning approaches. To mitigate these challenges, this study introduces a Reinforced Feature and Multiscale Feature Fusion Network (RFMFFNet) tailored for semantic interpretation of soil elements. The model incorporates a rectangular calibration attention (RCA) module into a ResNet101 backbone to recalibrate feature responses in critical regions, thereby improving scale adaptability and the preservation of fine geological structures. A complementary multiscale feature fusion (MFF) component is further designed by combining sparse self-attention with pyramid pooling, enabling richer context aggregation while reducing computational redundancy. Comprehensive experiments on the Landsat-8 and Sentinel-2 datasets verify the effectiveness of the proposed framework. RFMFFNet consistently achieves superior segmentation performance compared with several mainstream deep learning models. On the Landsat-8 dataset, the oPA and mIoU increase by 2.4% and 2.6%, respectively; on the Sentinel-2 dataset, the corresponding improvements reach 4.3% and 4.1%.

Contextual modulation of language comprehension in a dynamic neural model of lexical meaning.

The fashion industry increasingly relies on artificial intelligence technologies to enhance creative workflows and accelerate design innovation. This research presents a comprehensive framework that employs Generative Adversarial Net- works and advanced diffusion models to generate high-quality fashion imagery from textual descriptions. The proposed system integrates Stable Diffusion architecture with specialized text preprocessing pipelines to create diverse, photo realistic fashion designs that align with textual specifications while maintaining aesthetic coherence and commercial viability. The framework was evaluated using a dataset of 10,000 high-resolution fashion images, with systematic assessment conducted across multiple performance dimensions including creativity, aesthetic appeal, design diversity, and semantic consistency. Experimental results demonstrate exceptional performance in creative design generation, achieving average scores of 4.7 for originality and 4.5 for aesthetic quality based on comprehensive evaluation by thirty participants. The system successfully produces varied design alternatives from similar prompts, indicating robust exploration of design possibilities rather than repetitive pattern generation. While text prompt accuracy achieved a moderate score of 3.8, highlighting opportunities for enhanced semantic interpretation, the overall results validate the framework’s capability to support professional fashion design workflows. The research contributes to the growing body of knowledge in AI-assisted creative applications and demonstrates significant potential for transforming traditional fashion design processes through intelligent automation and creative augmentation technologies.

From text to meaning: Semantic interpretation of non-standardized metadata in piping and instrumentation diagrams

The high growth in digital legal documents has brought about immense tasks in information retrieving, categorizing, and decision-making processes about the law. Manual legal analysis is still time-consuming, inconsistent and heavily resource-dependent. Thus, there is a requirement for a transition to automated LegalTech. The current study examines how legislative acts in the European Union (EU) can be predicted based on legal decisions made using Natural Language Processing (NLP) and hybrid deep learning. Based on the use of the EurLex corpus with more than 142,000 legislative texts, this study focuses on classifying three major classes: Directives, Regulations, and Decisions. We employ a multi-stage pipeline in order to deal with not only the super-ordinary linguistic diversity of the legal texts but also the imbalance between the masses and categories. Wepropose and test composite deep learning models that join the Convolutional Neural Networks (CNNs), the Long Short-term memory (LSTM) and the Gated Recurrent Unit (GRU). These have been improved by adding attention mechanisms in order to cover both long-distance and local phrase structures. Its semantical interpretation is also reinforced in the model through FastText embedding that captures the subword law vocabulary. The findings of our experiment suggest that hybrid modelsare more accurate, recall, and F1-score than the traditional baselines of machine learning, especially in terms of multi-class classification of the legislative. It is enhanced by the incorporation of structured metadata, including Act Type and Subject Matter, that adds meaningful value to feature representation and generates a higher level of interpretability. This study focused on legislative data, while ignoring judicial data. It contributes to the use of AI in legal analytics, making a case for a scalable andexplainable document classification approach that transcends linguistic and jurisdictional boundaries.

This research examines idiomatic expressions found in the humorous dialogues of Legally Blonde (2001) using Leech’s semantic theory. The study aims to identify the idioms used in the film, describe their meanings based on semantic interpretation, and explain how they contribute to humor. A descriptive qualitative method was applied, with the film script serving as the primary data source. The researcher carefully read the script, selected scenes containing humorous or informal conversations, and identified idioms spoken by the characters. Each idiom was then analyzed in context and categorized into four types: figurative, slang/colloquial, euphemistic, and hyperbolic. The findings reveal 17 idioms that support the film’s comedic tone. Slang and colloquial idioms appear most frequently, reflecting the informal and youthful atmosphere of the film. Figurative idioms express emotion and personality, euphemistic soften direct statements, and hyperbolic idioms create exaggerated effects that strengthen humor. Overall, the study shows that idioms not only function as linguistic expressions but also play an important role in developing character identity and enhancing the humorous style of the film

This study investigates why the Department of X(cross)-Cultural Studies at Kookmin University designates its field of emotional inquiry not as the “Science of Emotions” but as the “Science of Feelings.” Previous scholarship has typically approached emotion either as a measurable physiological reaction within psychology and neuroscience or as an object of moral, semantic, and ontological interpretation within philosophy and the humanities. Yet these approaches fall short of clarifying emotion as an integrated process in which bodily changes are mapped and consciously experienced by the mind. Accordingly, this research distinguishes “Emotion” as the physiological level of bodily alteration and “Feeling” as the conscious awareness through which such alteration becomes an experience of one’s own existence. It thereby argues for understanding emotion not merely as a physiological response but as an ontological mode of experience. Methodologically, the study first examines Antonio Damasio’s theory to demonstrate that Feeling underlies judgment, decision-making, and the emergence of selfhood. Second, it analyzes Spinoza’s account of affect as the conjunction of a bodily modification and the idea of that modification. Third, it considers Toegye Yi Hwang’s Seong–eong theory, in which human nature(seong) manifests as feeling(jeong) through the medium of gi, revealing a structure consistent with the integration of body and mind. The comparative findings show that all three perspectives conceptualize emotion as an existential event grounded in the convergence of bodily change and its conscious recognition. Consequently, the study argues that emotion functions as a foundational principle through which existence recognizes itself and forms relationships with the world, thereby serving as a generative source for cognition, ethics, artistic creation, and cultural formation. Defining emotional inquiry as the “Science of Feelings” is therefore academically justified, as it understands emotion as a structure of self-awareness and articulates a renewed direction for contemporary emotional studies.

The rapid advancement of Artificial Intelligence (AI) has transformed diverse fields of knowledge, including Islamic textual studies. Within this context, the integration of AI into Hadith scholarship presents new opportunities for automation, verification, and knowledge extraction, while simultaneously introducing epistemological and ethical challenges. This study aims to systematically map and analyze global research on the application of AI in Hadith studies, identifying dominant technologies, methodological trends, key challenges, and future research directions. Employing a Systematic Literature Review (SLR) based on the PRISMA framework, 19 Scopus-indexed studies published between 2013 and 2025 were analyzed to trace publication dynamics and methodological patterns. The results reveal growing scholarly attention since 2019, with research evolving from machine learning applications for Hadith classification toward deep learning, natural language processing (NLP), and transformer-based models. AI has been predominantly applied in three domains: classification, authentication through isnād and matn analysis, and semantic or textual interpretation. Despite notable progress, persistent limitations remain, including the absence of standardized benchmark datasets, limited explainability of AI models, and weak integration between algorithmic reasoning and Islamic epistemology. The study underscores the need for ethically grounded and explainable AI frameworks aligned with uṣūl al-ḥadīth principles and maqāṣid al-sharī‘ah values to ensure theological integrity and interpretive transparency. Conceptually, it contributes to defining Digital Hadith Science as an emerging interdisciplinary field bridging data science and Islamic scholarship. The paper concludes by outlining a forward-looking research agenda emphasizing multilingual data infrastructures, epistemologically informed AI design, and collaborative frameworks between computer scientists and Islamic scholars.

The research objective is to consider hyponymy (specification) as a type of lexical-semantic transformation in the translation of the Yakut heroic epic Olonkho into English. The object of the study was the olonkho “Дьулуруйар Ньургун Боотур” by Platon A. Oyunsky and its English translation “Nurgun Botur the Swift.” The main research methods are the comparative method and the method of semantic interpretation of dictionary definitions. As a result, an analysis of the use of hyponymy (specification) in translation of olonkho was carried out. Different types of hyponymic transformation are identified, and the reasons for its use in translation are determined. The results of this study can be used in the development of a theory of translation of the Yakut language, directly in the translation of olonkho, and also as additional material for the course of teaching the theory and practice of translation.

The object of the paper is Russian multiplicative verbs, that is verbs denoting repeating homogenous acts, cf. šagat’ («to step»), maxat’ («to wave»), ikat’ («to hiccup»), etc. My goal is to specify the definition of this class of verbs and to interpretate from the semantic point of view a well known systemic ambiguity of some sentences with such a verb. The study is carried out within the frame of Moscow semantic school. The problem is that there are many verbs which denote processes consisting of repeating homogenous acts, and still such verbs are never classified as multiplicative; cf. idti («to walk»); bežat’ («to run»); kosit’ travu («to mow the grass»), etc. The point is that in the focus of the meaning of a multiplicative verb are repeating homogenous acts while in the focus of idti ‘to walk’, bežat’ ‘to run’, kosit’ travu ‘to mow the grass’, etc. is purpose of action. I also demonstrate that meaning of many multiplicative verbs contains semantic component ‘once or more than once’, and because of that in some contexts they denote not series of homogenous acts but a single act. I call such multiplicative verbs non-strict as opposed to strict multiplicatives which always denote repeating homogenous acts (cf. drožat’ «to tremble for some time»). From this point of view I discuss aspectual pairs like šagat’ – šagnut’, maxat’ – maxnut’. The semantic interpretation of the verbs under discussion allows us to reveal the direction of derivation in the pair multiplicative – semelfactive.