Search Results (72)

Accounting manipulation undermines the integrity of financial reporting and can distort key performance indicators, yet its quantitative effects on accounting quality (AQ) and value-related metrics remain underexplored. This study analyses U.S. publicly traded firms involved in accounting manipulation between 2017 and 2019, comparing them with matched non-manipulative industry peers to assess differences in AQ. It also examines potential links between manipulation-related AQ distortions and changes in Economic Value Added (EVA), stock prices, trading volumes, and dividend payouts. The sample includes 57 manipulation-affected firms and 57 matched controls, identified through SEC enforcement filings and the Violation Tracker database. Financial and stock data were sourced from EDGAR, ORBIS, and Morningstar. AQ was measured using discretionary accruals estimated via the Kasznik model. Correlation analysis tested associations between AQ and the selected performance indicators. Results show that firms involved in accounting manipulations had significantly lower AQ than their peers. However, no consistent correlations were found between AQ and EVA, dividends, stock prices, or volumes during the manipulation period. These findings suggest that the performance effects of manipulations are case-specific and shaped by additional factors, underscoring the importance of strong regulatory oversight and high-quality accounting practices. Ethically, our evidence underscores that misreporting corrodes investor trust and the public-interest mandate of financial reporting; accordingly, we stress the duties of boards, executives, auditors, and regulators to uphold faithful representation and timely disclosure, and to remediate misreporting when detected. Full article

Conventional teacher assessment is labor-intensive and subjective. Prior LLM-based systems improve scale but rely on post hoc rationales and lack built-in trust controls. We propose an explainable-by-design framework that couples (i) Dual-Lens Hierarchical Attention—a global lens aligned to curriculum standards and a local lens aligned to subject-specific rubrics—with (ii) a Trust-Gated Inference module that combines Monte-Carlo-dropout calibration and adversarial debiasing, and (iii) an On-the-Spot Explanation generator that shares the same fused representation and predicted score used for decision making. Thus, explanations are decision-consistent and curriculum-anchored rather than retrofitted. On TeacherEval-2023, EdNet-Math, and MM-TBA, our model attains an Inter-Rater Consistency of 82.4%, Explanation Credibility of 0.78, Fairness Gap of 1.8%, and Expected Calibration Error of 0.032. Faithfulness is verified via attention-to-rubric alignment (78%) and counterfactual deletion tests, while trust gating reduces confidently wrong outputs and triggers reject-and-refer when uncertainty is high. The system retains 99.6% accuracy under cross-domain transfer and degrades only 4.1% with 15% ASR noise, reducing human review workload by 41%. This establishes a reproducible path to trustworthy and pedagogy-aligned LLMs for high-stakes educational evaluation. Full article

This article is an attempt to empirically establish a new category of social beliefs defined as postmodern beliefs. They are cognitive categorizations of social and media messages regarding ways of understanding the world which are based on the basic assumptions of postmodernism, quite widely recognised as fundamental. The theoretical model adopted in the article assumes the existence of three beliefs: antifundamentalism, absolutization of freedom and relativization of truth. The hypothesised concept was operationalized as Postmodern Beliefs Questionnaire (PMBQ). Verification studies were carried out on three groups of over 600 people. The verification of the tool was carried out by using exploratory factor analysis (EFA) to select the appropriate pool of statements, then data in two subsequent datasets was analysed using Confirmatory Factor Analysis (CFA) to empirically verify the selected set of statements and estimate relevant parameters. The tool constructed allows for investigating the distinguished beliefs at a satisfactory level of reliability and validity. It can be used to measure the extent to which the representations that make up the popular understanding of postmodernism have been recognised and built into the overall belief system about the world of the respondents. The distinguished postmodern beliefs differ in terms of relations with other social beliefs of the respondents, such as anthropocentrism, traditionalism, faith in a just world, as well as the attitude of individuals to material values or their individualistic orientation. Full article

This article presents an innovative methodological approach for the documentation and enhancement of ancient ironworking heritage in the Guéra region of Chad. By combining ethno-historical and archaeological surveys, 3D modeling with Blender, and the generation of images and video sequences through artificial intelligence (AI), we propose an integrated production pipeline enabling the faithful reconstruction of three types of metallurgical furnaces. Our method relies on rigorously collected field data to generate multiple and plausible representations from fragmentary information. A standardized evaluation grid makes it possible to assess the archaeological fidelity, cultural authenticity, and visual quality of the reconstructions, thereby limiting biases inherent to generative models. The results offer strong potential for integration into immersive environments, opening up perspectives in education, digital museology, and the virtual preservation of traditional ironworking knowledge. This work demonstrates the relevance of multimodal approaches in reconciling scientific rigor with engaging visual storytelling. Full article

This article explores the voices of sailors across time, focusing on how song and prayer animate the nautical cult of St. Nicholas of Myra from the Middle Ages to the present. Drawing on hagiography, poetry, and music, it examines how medieval sources portray sailors’ cries to St. Nicholas during storms at sea, often depicting univocal, affective pleas that provoke divine response. These representations—especially in Latin sequences such as Congaudentes exultemus—highlight the cultural weight of the literal and metaphorical voice within miracle narratives. The article then bridges medieval and modern devotional soundscapes through nineteenth- and twentieth-century ethnographic collections from Apulia, Italy, particularly through the work of folklorists Saverio La Sorsa and Alfredo Giovine. Their records of Barese sailors’ songs and prayers to St. Nicholas—still sung today—provide embodied counterpoints to the mediated voices of medieval texts. Through this transhistorical lens, I argue that voice operates as connective tissue in the devotional lives of seafarers: an expression of fear, faith, and communal identity. By amplifying sailors’ voices in text, song, and performance, both medieval and modern traditions construct a vivid aural archive that affirms the enduring relationship between St. Nicholas and those who navigate the dangers of the sea. Full article

Information geometry provides a data-informed geometric lens for understanding data or physical systems, treating data or physical states as points on statistical manifolds endowed with information metrics, such as the Fisher information. Building on this foundation, we develop a robust mathematical framework for analyzing data residing on Riemannian manifolds, integrating geometric insights into information-theoretic principles to reveal how information is structured by curvature and nonlinear manifold geometry. Central to our approach are tools that respect intrinsic geometry: gradient flow lines, exponential and logarithmic maps, and kernel-based principal component analysis. These ingredients enable faithful, low-dimensional representations and insightful visualization of complex data, capturing both local and global relationships that are critical for interpreting physical phenomena, ranging from microscopic to cosmological scales. This framework may elucidate how information manifests in physical systems and how informational principles may constrain or shape dynamical laws. Ultimately, this could lead to groundbreaking discoveries and significant advancements that reshape our understanding of reality itself. Full article

This study investigates Weibull distribution modeling for data under grouped observations. Two data grouping methods (equal-width and optimal) were compared for estimating parameters of the Weibull distribution using maximum likelihood estimation (MLE) in each case. Methodologically, our contribution is twofold: First, we derive the correct Fisher information matrix for grouped data in the two-parameter Weibull and use it to compute optimal interval boundaries. Second, we derive maximum likelihood estimators for data grouped under these optimal intervals. The fit of the assumed distributions was evaluated using chi-squared goodness-of-fit tests. We also calculated Asymptotic Relative Efficiency (ARE) to compare the precision of parameter estimates across different grouping approaches. Optimal boundaries yielded systematically higher ARE than equal-width grouping in 100% of comparisons for the shape parameter $c$. Gains for the scale parameter $b$ were smaller and occurred in about 62% of cases. Optimal grouping also produced generally higher chi-squared $\left({\chi }^2\right)$ goodness-of-fit p-values than equal-width grouping, indicating a better fit. From a symmetry standpoint, the Weibull distribution is inherently asymmetric, with the degree of asymmetry governed by the shape parameter $c$. We show that the choice of grouping affects the estimate of $c$ and, thus, the inferred skewness, further explaining why optimally designed intervals yield both higher precision and a more faithful representation of failure behavior. Full article

I use Shakespeare to teach acting to students. A key to my work is impression management: what Shakespeare called reputation. I view the management of reputation as a route into Shakespearean character, which I present to students as a mask attuned to sacred values. The physical basis from which the actor can discover the mask is what Hamlet calls ‘smoothness’, which I explain with an acting exercise. We discover the force of sacred values by noticing the ubiquity of keywords in the text such as honor, virtue, reason, shame and faith. By holding characters to the fire of their sacred values, I shift the actor’s attention from an individualist idea of authentic representation towards a sense of character as a battleground of mind-shaping. The resulting performance work is scaled up to a more expansive and energized degree than the actor may be used to delivering in a social media-saturated environment in which what is often prioritized is a quasi-confessional self-revelation. The revelation of an inner life then emerges through a committed exploration of antithetical relations, a strategy basic both to mask work and to Shakespeare’s poetics. The actor finds their personal connection to the material by facing the contradiction between the objective standards of behavior demanded of the character and the character’s attempt to control their status, that is, how they are seen. The final value of the performance work is that the actor learns how to manage their reputation so that they come to appear like a giant who is seen from a distance. Full article

Underwater inspection images of hydraulic structures often suffer from haze, severe color distortion, low contrast, and blurred textures, impairing the accuracy of automated crack, spalling, and corrosion detection. However, many existing enhancement methods fail to preserve structural details and suppress noise in turbid environments. To address these limitations, we propose a compact image enhancement framework called Wavelet Fusion with Sobel-based Weighting (WWSF). This method first corrects global color and luminance distributions using multiscale Retinex and gamma mapping, followed by local contrast enhancement via CLAHE in the L channel of the CIELAB color space. Two preliminarily corrected images are decomposed using discrete wavelet transform (DWT); low-frequency bands are fused based on maximum energy, while high-frequency bands are adaptively weighted by Sobel edge energy to highlight structural features and suppress background noise. The enhanced image is reconstructed via inverse DWT. Experiments on real-world sluice gate datasets demonstrate that WWSF outperforms six state-of-the-art methods, achieving the highest scores on UIQM and AG while remaining competitive on entropy (EN). Moreover, the method retains strong robustness under high turbidity conditions (T ≥ 35 NTU), producing sharper edges, more faithful color representation, and improved texture clarity. These results indicate that WWSF is an effective preprocessing tool for downstream tasks such as segmentation, defect classification, and condition assessment of hydraulic infrastructure in complex underwater environments. Full article

Explainable Artificial Intelligence (XAI) is increasingly important in computer vision, aiming to connect complex model outputs with human understanding. This review provides a focused comparative analysis of representative XAI methods in four main categories, attribution-based, activation-based, perturbation-based, and transformer-based approaches, selected from a broader literature landscape. Attribution-based methods like Grad-CAM highlight key input regions using gradients and feature activation. Activation-based methods analyze the responses of internal neurons or feature maps to identify which parts of the input activate specific layers or units, helping to reveal hierarchical feature representations. Perturbation-based techniques, such as RISE, assess feature importance through input modifications without accessing internal model details. Transformer-based methods, which use self-attention, offer global interpretability by tracing information flow across layers. We evaluate these methods using metrics such as faithfulness, localization accuracy, efficiency, and overlap with medical annotations. We also propose a hierarchical taxonomy to classify these methods, reflecting the diversity of XAI techniques. Results show that RISE has the highest faithfulness but is computationally expensive, limiting its use in real-time scenarios. Transformer-based methods perform well in medical imaging, with high IoU scores, though interpreting attention maps requires care. These findings emphasize the need for context-aware evaluation and hybrid XAI methods balancing interpretability and efficiency. The review ends by discussing ethical and practical challenges, stressing the need for standard benchmarks and domain-specific tuning. Full article

The Yogācārabhūmi, compiled by Saṅgharakṣa, was first introduced to China by An Shigao’s abridged translation (T607, Daodi jing 道地經), later, in 284 CE, Dharmarakṣa produced a more comprehensive version (T606, Xiuxing daodi jing 修行道地經). Lacking extant Sanskrit or Pali parallels, the text is difficult to interpret literally, and the differences between T607 and T606 add to the analytical challenges. However, a substantial section in both translations describing omens of impending death in the sick exhibits systematic parallels with Indian Āyurvedic texts, such as the Caraka-saṃhitā and Suśruta-saṃhitā. These parallels help clarify the ambiguous passages through comparative analysis. This study explores the translation strategies of An Shigao and Dharmarakṣa in introducing Indian medical concepts to China. An Shigao adopted a localization strategy, replacing foreign terms with analogous Chinese concepts. His terminology, corroborated by usage in Eastern Han or earlier Chinese texts—particularly excavated manuscripts—supports claims in the Chu sanzang ji ji regarding his expertise in medicine and divination. By contrast, Dharmarakṣa’s Xiuxing daodi jing sought greater fidelity to the Indian source material, offering a more detailed and systematic presentation of Āyurvedic knowledge. However, Dharmarakṣa did not entirely abandon An Shigao’s localization approach. He adopted a balanced strategy that combined faithful representation with cultural adaptation, reflecting the broader capacity of his more diverse and sophisticated audience to engage with complex and extensive foreign knowledge. Full article

Retrieval-Augmented Generation (RAG) has become a central approach to enhance the factual consistency and domain specificity of large language models (LLMs) by incorporating external context at inference time. However, most existing RAG systems rely on dense vector-based similarity, which fails to capture complex semantic structures, relational dependencies, and multimodal content. In this paper, we introduce Graph Codes—a matrix-based encoding of Multimedia Feature Graphs—as an alternative retrieval paradigm. Graph Codes preserve semantic topology by explicitly encoding entities and their typed relationships from multimodal documents, enabling structure-aware and interpretable retrieval. We evaluate our system in two domains: multimodal scene understanding (200 annotated image-question pairs) and clinical question answering (150 real-world medical queries with 10,000 structured knowledge snippets). Results show that our method outperforms dense retrieval baselines in precision (+9–15%), reduces hallucination rates by over 30%, and yields higher expert-rated answer quality. Theoretically, this work demonstrates that symbolic similarity over typed semantic graphs provides a more faithful alignment mechanism than latent embeddings. Practically, it enables interpretable, modality-agnostic retrieval pipelines deployable in high-stakes domains such as medicine or law. We conclude that Graph Code-based RAG bridges the gap between structured knowledge representation and neural generation, offering a robust and explainable alternative to existing approaches. Full article

Multi-Hop Knowledge Graph Question Answering (multi-hop KGQA) aims to obtain answers by analyzing the semantics of natural language questions and performing multi-step reasoning across multiple entities and relations in knowledge graphs. Traditional embedding-based methods map natural language questions and knowledge graphs into vector spaces for answer matching through vector operations. While these approaches have improved model performance, they face two critical challenges: the lack of clear interpretability caused by implicit reasoning mechanisms, and the semantic gap between natural language queries and structured knowledge representations. This study proposes the DRKG (Decomposed Reasoning over Knowledge Graph), a constrained multi-hop reasoning framework based on large language models (LLMs) that introduces explicit reasoning plans as logical boundary controllers. The innovation of the DRKG lies in two key aspects: First, the DRKG generates hop-constrained reasoning plans through semantic parsing based on LLMs, explicitly defining the traversal path length and entity-retrieval logic in knowledge graphs. Second, the DRKG conducts selective retrieval during knowledge graph traversal based on these reasoning plans, ensuring faithfulness to structured knowledge. We evaluate the DRKG on four datasets, and the experimental results demonstrate that the DRKG achieves 1%–5% accuracy improvements over the best baseline models. Additional ablation studies verify the effectiveness of explicit reasoning plans in enhancing interpretability while constraining path divergence. A reliability analysis further examines the impact of different parameters combinations on the DRKG’s performance. Full article

This paper explores the possibility of counteracting the crisis of culture and institutions by investing in the identity values of the user-actor within digital spaces built for the purpose. The strategy is applied to the analysis of three Catalan cloisters (Spain), with a focus on the representation of the cloister of Sant Cugat (Barcelona). Heuristic picklocks are found in the semantic richness proposed by Marius Schneider exclusively on the verbal level. The authors interpret the contents and transcribe them into graphic signs and digital denotations of sounds and colors. They organize proprietary ontologies, or syntagmatic lines, to be entrusted to the management of computer algorithms. The syncretic culture that characterized the medieval era allowed the ability to mediate science and faith to be entrusted to the mind of the praying monk alone in every canonical hour. The hypothesis that a careful direction has programmed the ways in which to orient souls to “navigate by sight” urges the authors to find the criteria that advanced statistics imitates to make automatic data processing “Intelligent”. In step with the times and in line with the most recent directions for the Safeguarding of Heritage, the musical, astral, and narrative rhythms feared by Schneider are used to inform representative models, to increase not only the visual perception of the user (XR Extended Reality) but also to solicit new analogies and illuminating associations. The results return a vision of the culture of the time suitable for shortening the distances between present and past, attracting the visitor and, with him, the resources necessary to protect and enhance the spaces of the Romanesque era. The methodology goes beyond the contingent aspect by encouraging the ‘remediation’ of contents with the help of machine learning. Full article

Transformer models produce advanced text representations that have been used to break through the hard challenge of natural language understanding. Using the Transformer’s attention mechanism, which acts as a language learning memory, trained on tens of billions of words, a word sense disambiguation (WSD) algorithm can now construct a more faithful vectorial representation of the context of a word to be disambiguated. Working with a set of 34 lemmas of nouns, verbs, adjectives and adverbs selected from the National Reference Corpus of Romanian (CoRoLa), we show that using BERT’s attention heads at all hidden layers, we can devise contextual vectors of the target lemma that produce better clusters of lemma’s senses than the ones obtained with standard BERT embeddings. If we automatically translate the Romanian example sentences of the target lemma into English, we show that we can reliably infer the number of senses with which the target lemma appears in the CoRoLa. We also describe an unsupervised WSD algorithm that, using a Romanian BERT model and a few example sentences of the target lemma’s senses, can label the Romanian induced sense clusters with the appropriate sense labels, with an average accuracy of 64%. Full article