Search Results (570)

Animal micro-expressions constitute transient behavioral windows that link internal states to externally observable signals, while artificial intelligence (AI) serves as the critical bridge that transforms these windows into measurable, interpretable, and applicable scientific tools. Rather than imposing a human-centered lexicon of expressions, AI-driven decoding aims to develop biologically grounded and increasingly comparable behavioral biomarkers that link computable facial dynamics to internal states; however, a validated universal cross-species framework has not yet been established. This review summarizes the common behavioral characteristics of animal micro-expressions, their cross-species expressive forms, and functional differences; systematically outlines the methodological spectrum through which AI captures, encodes, recognizes, and interprets these brief yet complex signals; and finally discusses the expanded applications of AI plus micro-expression analysis in basic research, clinical diagnosis, and animal welfare governance, thereby promoting a paradigm shift in the decoding of animal micro-expressions. Full article

The Greek dialects of Lesvos and Adrianople (present-day Edirne) during the time of the Ottoman Empire have much in common. Both are northern dialects and show considerable influence from Turkish; moreover, both were documented in the early 20th century, before the fall of the Ottoman Empire, by Kretschmer, in 1905, for Lesvos and by Ronzevalle, in 1911, for Adrianople. A comparative look at the two dialects reveals deep penetration of Turkish into their respective lexicons, with consequences for their phonology, and for Adrianople, into aspects of its grammar too. I focus here on a phonological matter, namely the conditions under which voiced stops (D) occur. Kretschmer’s song recordings from Lesvos show D only after full nasals (N), i.e., in ND clusters, while later documentation, Newton 1972, shows no nasality, with only D as the realization, even in words with ND earlier. Adrianople Greek shows intervocalic D but also ND clusters, only in Turkish loans and always matching the consonantism of the Turkish source; such is the case too in ND/D variation in rebetika songs from Istanbul. This suggests that the issue really involves a kind of dialect mixing, and it is argued that this holds even for Lesvos, and even for sociolinguistically coded variation. Thus, bringing Adrianople Greek into consideration clarifies the nature of voiced stop realizations in Lesvos phonology. Full article

As tourism shifts from sightseeing to experience-oriented consumption, understanding how tourist perception differs across heterogeneous destination types and spatial scales remains challenging. Using 23,439 Volunteered Geographic Information (VGI) reviews archived for six destinations in Shangrao, China, this study compares mountain-type and rural comprehensive destination products. These are operational dominant-function categories rather than mutually exclusive geomorphological classes. The archive supports fine-grained sentiment, topic, and semantic-network analyses; annual temporal comparisons use the full 23,439-review corpus covering 2019–2025, whereas a separate subset of reviews posted from 1 August 2022 with official IP labels, aggregated into 2022–2024 province–year observations, supports Pooled Ordinary Least Squares (Pooled OLS) estimation. A hybrid lexicon–XLM-RoBERTa workflow, BERTopic, semantic co-occurrence analysis, and Pooled OLS are integrated in a cross-scale framework. Static results reveal shared strengths and weaknesses—high scenery and overall-experience evaluations but low price evaluations—alongside type-specific structures: mountain reviews concentrate on natural scenery, climbing effort, and accessibility, whereas rural reviews span village landscapes, cultural activities, accommodation, and nighttime experiences. Temporally, mountain demand retains a stable scenic core while accessibility concerns become more salient; rural demand shifts from traditional agricultural landscapes toward nighttime performances and other experience-oriented products. Cross-scale regressions identify destination- and dimension-specific correlates rather than causal drivers: urbanization is positively associated with several rural evaluations, while ecological contrast, climatic difference, and competing scenic resources are associated with more critical assessments in selected dimensions. The findings show that perception differences arise from the interaction of destination product structures and origin-region contexts, supporting differentiated accessibility management for mountain destinations and balanced product innovation, service improvement, and commercialization control for rural destinations. Full article

The integration of green finance and digital technology exhibits considerable methodological diversity; however, existing scholarship lacks a systematic account of how empirical evidence is generated in this field. This scoping (PRISMA-ScR) review develops a literature classification framework centered on methodological approaches and applies it to 423 English-language articles on the green finance–digital technology nexus from the Web of Science Core Collection (2018 to April 2026). Research practices are organized along a two-dimensional framework spanning the epistemological objectives (explanatory vs. predictive) and the technical toolkit (traditional econometrics vs. machine learning), with causal machine learning integrating both paradigms. Each study is assigned a primary methodological label via a lexicon-based multi-label classification engine employing a 2 × 2 weight matrix. Weight sensitivity analysis confirms that classification outcomes are robust to parameter perturbation. Inter-coder reliability was near-perfect (Cohen’s κ = 0.93); algorithm–gold standard agreement was substantial (κ = 0.76, F1 = 0.79). The classification reveals that traditional econometrics remains the most prevalent method (26.95%), followed by predictive machine learning (17.26%) and technology application (16.31%), while causal machine learning (9.69%) remains underutilized and behavioral research is virtually absent (0.95%). Over a quarter of studies (26.24%) employ multiple methodologies. These findings provide a methodological map for researchers and policymakers navigating the green finance–digital technology landscape. Full article

Toxic language detection is an important task for online content moderation. However, existing models often associate neutral group-related terms—such as identity, regional, or demographic expressions—with toxic labels, producing false-positive bias in which benign texts containing such terms are misclassified as toxic. To address this, we propose NISC, a neuron-level iterative debiasing and semantic compensation framework for Chinese toxic language detection. Starting from a trained detector, NISC first identifies bias sources from high-confidence false-positive samples without a manually predefined lexicon. It then builds paired samples with and without each bias source and compares neuron-level toxic contributions between them. Guided by this contribution difference, NISC locates neurons that drive false-positive predictions and weakens them through iterative pruning. To offset the semantic degradation that pruning may cause, NISC uses a large language model to generate benign non-toxic samples that contain the identified bias sources but express no attack, insult, discrimination, or hatred, and applies them for lightweight compensation training with the pruning mask fixed. Experiments on multiple Chinese toxic language datasets show that NISC lowers the false-positive rate on group-related neutral texts while largely preserving conventional detection performance. Full article

Automatic extraction of genealogical information from historical archival-genealogical documents in Uzbek is an understudied problem for low-resource languages. Multi-layer NLP benchmarks are not sufficient to automatically identify individuals, family relationships, dates, place names, and archival identifiers in such texts. Also, the same people are mentioned in various forms: full name, pronoun (18.8%), initial, surname-name order, indirect expression (9.4%), and title. Existing NER and relation extraction corpora are mainly focused on high-resource languages or general domain texts and do not sufficiently cover the FAMILY_ROLE signals, historical spelling variants, and fond–opis–delos identifiers specific to Uzbek archival-genealogical texts. Proposed resource: We present the ArchiveGene Corpus, a controlled, fully synthetic, and reproducible five-layer resource consisting of 1000 Uzbek archival-genealogical-style documents, divided into 700 training, 150 validation, and 150 test documents. The corpus contains 8366 named entities, 10,625 person mentions, 2000 coreference chains, and 1000 genealogical relation triples. The dataset was generated using a deterministic template-based pipeline and a lexicon of Uzbek names, and is fully reproducible. Inter-annotator agreement values were 0.847 for NER, 0.793 for coreference, and 0.821 for RE, according to Cohen’s κ. Comparative results are presented with four baseline models (rule-based, BiLSTM-CRF, mBERT, and XLM-RoBERTa). The dataset is openly hosted on the Zenodo platform under the CC BY 4.0 license; concept DOI: 10.5281/zenodo.20670360, v1.1.1 version DOI: 10.5281/zenodo. 21429998. Scientific significance: To the best of our knowledge, ArchiveGene is among the first openly released, controlled synthetic resources for Uzbek that integrates named-entity recognition, person-mention detection, coreference resolution, genealogical relation extraction, and final tuple generation within a single annotation framework. The baseline analysis provides three main conclusions: (1) on the clean synthetic test set, the transformer models already reach 100.00 Micro-F1 for NER and 100.00 Macro-F1 for coreference-aware relation extraction, so coreference aggregation adds little on synthetic data (+2.25 for mBERT and +0.04 for XLM-RoBERTa) but its contribution is expected to grow on real archival text; (2) the rule-based and heuristic baselines lag far behind (Macro-F1 50.28 and 70.73) and fail entirely on spouse_of, showing the limits of lexical rules; and (3) a zero-shot evaluation on a real-document pilot reduces NER Micro-F1 from 100.00 to 22.17, indicating that the synthetic corpus is trivially learnable and that real-archival validation is essential. Full article

Ideathons and pitch competitions have expanded rapidly as experiential learning devices, but the textual artefacts they produce—presentation slides—remain under-examined. This study applies interpretable computational text analysis to 104 ideathon decks (1289 content slides) from four cohorts: Cambodia 2023, 2024, 2025 and Thailand 2025. Measures include lexical frequency, TF-IDF, lexicon-based sentiment, a ten-component pitch-completeness proxy, numerical density, and Jaccard similarity. Because sector designation was absent in Cambodia 2023 and present from 2024 onward, the longitudinal Cambodian data support an observational cohort comparison with an institutional change between cohorts; year effects, programme evolution, and sector designation cannot be separated. The Cambodia 2025 vs. Thailand 2025 contrast is a single-year cross-country comparison, not a longitudinal one. Between Cambodia 2023 and 2024, presentations show large Cohen’s d differences with 95% bootstrap confidence intervals (CIs) in total words, unique words, slide count, pitch completeness, and market-related vocabulary, alongside a small decline in type–token ratio. At fixed sector composition, Cambodia 2025 and Thailand 2025 differ sharply in surface vocabulary (top-50 Jaccard = 0.176): Cambodia leans toward agriculture, rural markets, and community development, while Thailand leans toward AI, learning, and cassava-centric agronomy. AI use was not directly measured, so all claims about generative AI are hypothesis-generating; the drop in within-cohort pairwise Jaccard from 2024 to 2025 (0.061 → 0.041) is consistent with—but does not establish—an augmentative rather than homogenising effect of AI assistance. Findings are reported as descriptive associations and interpreted through the lens of constraint-based creativity and institutional theory. We discuss implications for curriculum designers who wish to balance structural templates with exercises that promote diverse problem framings. Full article

Sentiment analysis for Greek social media texts remains challenging because of limited annotated resources, linguistic variation, and domain-dependent sentiment expression. This study presents a topic-aware, lexicon-guided framework for sentiment classification across five reference domains in Greek social media. Domain-specific sentiment lexicons are activated according to the relevant domain and transformed into a structured representation comprising a token-level multi-channel lexical matrix and aggregate lexical descriptors. A fusion convolutional neural network combines these complementary components to classify sentiment while retaining explicit lexical evidence for inspection. The evaluation follows a leakage-free protocol: lexicons are constructed exclusively from the sentiment inner-training subset, validation data are used for model selection, and a held-out test set is reserved for final evaluation. The proposed fusion CNN achieved the strongest held-out sentiment result among the evaluated models, with an Accuracy of 0.8029 and a Macro-F1 of 0.7883, exceeding TF–IDF + Linear SVM and fine-tuned GreekBERT baselines in the present experimental setting. Ablation results show that the token-level lexical matrix and global descriptors provide complementary information. For domain routing, GreekBERT late fusion achieved an Accuracy of 0.9162 and a Macro-F1 of 0.9116. When lexicon activation used predicted rather than reference domains, the end-to-end sentiment pipeline achieved a Macro-F1 of 0.7569. These findings indicate that explicit domain-specific lexical knowledge can support an interpretable sentiment representation while making the effects of lexical coverage and topic-routing uncertainty visible. Full article

Background: Social-media-based risk screening is promising, yet deployment quality is constrained by noisy language, class imbalance, low explainability, and uncertainty handling. Objective: This work develops an application-specific BiLSTM-centered proxy risk screening framework that combines chaos-regularized learning, SMOTE balancing, SHAP explanation, and TF-IDF similarity fallback. Methods: Experiments used an annotated Twitter corpus (>31,000 English posts) with neutral, hate, and offensive labels, where hate/offensive content was treated only as a proxy risk indicator rather than as evidence of depression, self-harm, or clinical psychological distress. The workflow covered text normalization, tokenization, fixed-length sequence modeling, imbalance correction, and thresholded inference. The model stack used GloVe embeddings with recurrent layers and chaos-based dropout, trained with Adam plus adaptive stopping/scheduling. Operational safety was supported through lexicon checks and a balanced TF-IDF reference set for low-confidence cases. Results: Validation performance reached an F1-score of 0.86, accuracy 0.79, and AUC 0.93 under the reported pipeline. Component-level evidence further indicated that SHAP, thresholding, TF-IDF retrieval, and lexicon checks add explanation, uncertainty handling, reference context, and conservative escalation beyond the underlying classifier, although their marginal effects were not isolated through rerun ablation. Conclusion: The proposed system offers a practical tradeoff between predictive quality, interpretability, and safety-aware operation for proxy risk screening on noisy social media. The contribution is positioned as an integrated application pipeline rather than a new NLP algorithm, and future work should validate the framework with clinically grounded labels, leakage-safe resampling, ablation studies, and transformer baselines. Full article

Fake social media profiles increasingly resemble ordinary accounts, combining believable images, fluent biographies, copied posts, and selective engagement, which makes single signal detection unreliable. This study develops and evaluates a forensic natural language processing (NLP) framework that treats profile text as digital evidence rather than raw model input. Profiles were collected from Facebook, Instagram, X, Truth Social, and LinkedIn through network expansion and vocabulary guided scraping, then anonymized, merged, cleaned, and organized into structured text and profile records. Ground truth came from human eye annotation (referred to throughout as human eye review, human review, or manual review) supported by three LLM reviewers (referred to throughout as LLM assisted review), OpenAI GPT-4o, Anthropic Claude 3.5 Sonnet, and Google Gemini 1.5 Pro, with strict agreement logic producing confidence-tiered labels across platform-specific label spaces covering scam, fraud, harmful, and legitimate behavior. Each model received a defined set of label categories, was instructed to return a single label with a one sentence rationale, and worked independently with no cross-model communication. Multidimensional features, including TF-IDF representations, sentence embeddings, scam lexicon counts, sentiment and emotion scores, and behavioral indicators, were extracted through a pipeline built to keep links, hashtags, contact markers, and repeated phrasing as forensic signals rather than noise. Traditional classifiers, transformer models, and ensembles were trained on verified labels, and the best model per platform was applied to the full dataset of over 309,108 records under confidence filtered classification, with predictions below a 0.70 probability threshold flagged for human review. The framework links classification outputs to textual indicators through explainable AI, achieving weighted F1 scores between 0.86 and 0.96 across five platforms. These results show that profile text alone provides strong, production-relevant evidence for detecting fake and harmful accounts, even though the framework treats this as one evidence stream to be weighed alongside image and behavioral signals rather than as a final determination on its own. The accompanying evidence trail means an investigator or a trust and safety team can see why a profile was flagged rather than being handed a bare score, which is what turns a classifier into a tool that can be checked, challenged, and used directly in an investigation. Full article

This article examines the extent to which it is possible to reconstruct part of the technical knowledge employed in pre-Hispanic Andean constructions. Drawing on chronicles, colonial representations, linguistic studies, ethnography and direct architectural analysis, it demonstrates that Andean societies possessed specialised trades and a wider repertoire of tools than is commonly recognised. A central finding is the identification of gaveras—wooden moulds for making adobe bricks—whose recognition here shows that carpentry techniques were more developed than has been assumed. The poor documentation of these objects—several of them miscatalogued in Peruvian museums—has reinforced the mistaken assumption that complex construction tools were not used in ancient Peru. The study combines three approaches: direct observation of the built fabric (marks, imprints and textures that reveal construction processes), review of the Quechua vocabulary associated with tools and trades—drawn from three sixteenth- and seventeenth-century dictionaries consulted in the Langas repository of the CNRS—and comparison with current practices in Andean communities. This is complemented by a comparison with records from other cultures and a systematic search of the database of the Ministry of Culture of Peru, where plumb bobs, floats, shovels, axes, chisels and other instruments rarely analysed from a constructive perspective were located. Taken together, the article demonstrates that pre-Hispanic architecture was neither intuitive nor improvised, but the outcome of a technical tradition carried out by specialists, tools that are little known today, and construction methods based on measurement, alignment control, mortar transfer and careful surface finishing. Full article

In wine sensory evaluation, the integration of advanced data analysis techniques with more traditional approaches is essential for addressing challenges and improving practical applications. The current work explores some innovative text mining approaches for obtaining sensory information from various sources. One aspect of interest is creating specialised dictionaries and lexicons through automated processing (e.g., Natural Language Processing/NLP), so that raw text data can be converted into standardised sensory terms understood by specialists. Programmatic processing, in this context, refers to the automated extraction, transformation, consolidation, and analysis of sensory data, significantly reducing human error and allowing for the efficient handling of large data volumes, which is something not possible through manual processing only. The approach is illustrated using diverse data sources, such as structured database-type (for example, from wine sellers) and open-source data from sensory and consumer research publications. The manuscript introduces new ways to draw insights using coupled pipelines for dimension reduction (Multiple Correspondence Analysis/MCA), cluster analysis (Latent Semantic Analysis/LSA), and visual tools (network analysis, chord diagrams, wordvenn diagrams). This manuscript aims to provide practical tools and new insights for sensory specialists looking to enhance their evaluation techniques using text data, showcasing the versatility of advanced analysis techniques in extracting relevant sensory information. These tools use open-source languages such as Python and R and can be further applied for different product spaces. Full article

In this paper we study how external affective information should be integrated into a compact transformer-based text classifier. Rather than treating affective features as signals to be appended directly to the representation, we examine whether their contribution should be controlled through lightweight fusion mechanisms. The comparison focuses on scalar-gated fusion versus plain concatenation, using DistilBERT as the textual backbone and four affective resources: the NRC VAD Lexicon, VAD-BERT, Ekman-style emotion scores, and SenticNet. The evaluation is conducted on two English corpora with different label structures: a seven-class MentalHealth dataset and the fine-grained GoEmotions benchmark. Across both corpora, scalar gating consistently matches or outperforms concatenation in terms of Macro-F1. On MentalHealth, scalar gating improves all directly comparable configurations. On GoEmotions, it achieves the best overall Macro-F1 and improves most matched comparisons. Beyond static feature integration, we introduce affective flow (EmoFlow) representations derived from VAD-BERT, which model the evolution of valence, arousal, and dominance across segments of a text. These dynamic representations do not surpass the strongest static lexical resources in absolute performance, but they provide consistent improvements within the VAD-BERT family, particularly when combined with scalar gating or cross-attention. Our contribution is twofold. First, we show that a lightweight scalar gate provides an effective and interpretable mechanism for adaptively integrating low-dimensional affective side information into transformer-based classifiers. Second, we introduce affective flow representations that explicitly model how affect evolves within a document, enabling the analysis of both adaptive resource selection and intra-document affective dynamics. Together, these results suggest that the key issue is not only which affective resources to use, but also when and how they should influence the model. Full article

The renowned Sufi, commentator and Qurʾānīc exegete ʿAbdul Razzāq Kāshānī (d. 735 H/1335?, also pronounced as al-Qāshānī or Qāsānī) played a major role in disseminating the teachings of Ibn ʿArabī (d. 638 H/1240) in Ilkhanid (654 H/1256–750/1353), Iran. By teaching and training students such as Sharaf al-Dīn Dāwūd Qayṣarī (d. 751 H/1350), writing tirelessly on the main elements of Akbarīan mysticism, including waḥdat al-wujūd (oneness of being), wilāya (guardianship, sainthood), and khatm al-wilāya (the sealing of the sainthood), as well as commenting on Fuṣūṣ al-ḥikam (the Bezels of Wisdom), Kāshānī proved to be a true heir to the legacy of his late master. However, the account of his contribution to this tradition, which was quite new in Iranian milieu at that time, will not be complete if we neglect to mention his initiative in systematizing and theorizing Ibn ʿArabī’s teachings through his complex and detailed lexicons, including Muʿjam iṣṭilāḥāt al-ṣūfīya, Rashḥ al-zulāl fī sharḥ al-alfāẓ al-mutadāwilat bayn arbāb al-adhwāq wa-al-aḥwāl, and Laṭāʾif al-iʿlām fī ishārāt ahl al-ilhām. By means of his lexicalization of the challenging language of Ibn ʿArabī, Kāshānī facilitated a gain in popularity of the former’s teachings, as well as the widespread usage of those teachings both among Sufi circles and in the Ilkhanid court. Full article

As urban renewal shifts toward existing stock, historic-district regeneration requires evaluation methods that address material conditions, governance, and lived experience. Conventional physical metrics often miss subjective perceptions of access, management, and lived experience. Drawing on Lefebvre’s spatial triad, this study develops an operational, case-specific framework using user-generated content from Google Maps and Xiaohongshu. The analysis combines frozen BERT representations and K-means clustering for Gran Clariana, Barcelona (N = 876), with deterministic lexicon matching and corpus-specific sentiment procedures for two Shanghai historic districts (N = 522) and Het Westerpark, Amsterdam (N = 1310). In Shanghai, Spatial Practice was the most frequent dominant dimension, whereas Representational Space had the highest within-category positive sentiment share. For Gran Clariana, K = 3 was the most stable solution tested (mean ARI = 0.974), while Landscape Perceived Value was the leading profile discriminator (Welch F = 48.98, p < 0.001). In Westerpark, Spatial Practice was most lexically prominent, and Representational Space showed the largest, though weak, positive association with VADER sentiment (r = 0.138, p < 0.001). Taken together, these case-specific, exploratory mappings lack external validation but complement post-occupancy evaluation by interpreting platform-visible, inclusion-relevant perceptions rather than measuring demographic representation, affordability, accessibility compliance, or distributive justice. Full article