Search Results (155)

Grammatical error correction (GEC) for Chinese learner English is still dominated by sentence-level modeling, which limits discourse-level consistency and weakens adaptation to learner-specific error profiles. From an instructional perspective, these limitations also reduce the value of automated feedback as a basis for spaced-repetition instructional structures sensitive to individual differences. This study proposes a symmetry-aware dual-encoder architecture for context-aware GEC in Chinese learner English. A context encoder captures preceding-sentence information, while a source encoder integrates BERT-based semantic representations with Bi-GRU-based syntactic features for the current sentence. A gated decoder performs asymmetric fusion of local and contextual evidence. To better reflect corpus-level tendencies in Chinese learner English, a CLEC-informed augmentation strategy generates synthetic errors using empirical category frequencies as a coarse sampling prior. Experiments on CoNLL-2014, JFLEG, and CLEC show consistent improvements over strong neural baselines in F_0.5 and GLEU under the current desktop-oriented implementation setting. Nevertheless, the integration of BERT, dual encoders, and gated decoding introduces non-negligible computational overhead, and the present system is therefore better suited to desktop writing-support scenarios than to strict real-time or large-scale online deployment. The proposed framework thus provides a practical technical basis for personalized grammar feedback and for future spaced-repetition instructional designs in ESL writing support. Full article

Classroom humor is an important instructional strategy that enhances teaching effectiveness and improves student engagement. However, its automatic detection remains challenging due to the strong contextual dependency and implicit semantic shifts that characterize humorous expressions in teaching discourse. Conventional pretrained language models capture global semantics but often fail to focus on the subtle humor anchors that trigger incongruity. To address this issue, we propose T-HumorAGSA, a cognitive-inspired classroom teacher humor language detection model. The model employs BERT for contextualized semantic encoding, followed by a Gated Anchor-Guided Self-Attention (AGSA) mechanism that adaptively amplifies anchor-related features responsible for humor generation. A bidirectional gated recurrent unit (BiGRU) layer is further integrated to model long-range temporal dependencies within teaching utterances. T-HumorAGSA is evaluated on five datasets, including SemEval 2021 Task 7-1a, ColBERT, CCL2018, CCL2019 and the self-constructed teacher humor language dataset (T-Humor), demonstrating consistently strong performance. For instance, it achieves 0.9874 F1 on ColBERT and 0.9508 F1 on SemEval 2021 Task 7-1a, both outperforming the best baseline models. On the T-Humor dataset, the model attains a high F1 score of 0.9895, validating its capacity to detect subtle humorous cues in instructional discourse. The results demonstrate that the proposed model delivers excellent performance in classroom humor detection. Full article

Australian regional communities are actively seeking development pathways that generate local economic value while maintaining environmental and cultural integrity. In this context, GeoRegions have emerged in Australia as a community-led approach for recognising and interpreting geoheritage and associated abiotic–biotic–cultural (ABC) values through geotourism and geoeducation. The GeoRegion concept remains intentionally operationally flexible, but for regional communities encountering a myriad of barriers to sustainable geotourism implementation, any uncertainty for proponents about what constitutes an implementable GeoRegion and what resources and governance arrangements are required for credible and sustained delivery requires resolution. This study developed a stakeholder-informed conceptual model to clarify the practical ‘building blocks’ of GeoRegion establishment and the conditions under which GeoRegions can contribute to sustainability-oriented regional development. Using a design thinking framing and semi-structured interviews with thirteen expert participants, we used semantic discourse analysis to identify the factors perceived as essential to GeoRegion viability and legitimacy. We found that participants expected GeoRegions to be geologically centred, but their perceived value and long-term durability depend on (i) genuine community support and locally legitimate narratives (including Indigenous knowledge where appropriate), (ii) capable champions or coordinating groups, (iii) sustained resourcing for interpretation and visitor readiness, and (iv) a facilitative and not prescriptive role for government. Participants emphasised that GeoRegions should never be constrained by land tenure but cautioned that competing land uses, access logistics and uneven capacity across regions were highly influential in the delineation of feasible boundaries and management intensity. Our GeoRegion model differentiates core inputs (community mandate, knowledge co-production, geoheritage significance, human capacity and funding) from expected outputs (interpretive materials, geoeducation, geotourism, economic development, conservation outcomes and strengthened place identity), and we identify feedback that can either reinforce or erode sustainability outcomes over time. We argue that GeoRegions can provide a low-risk, scalable mechanism for geoconservation-informed regional development, particularly where formal protected-area tools or geopark ambitions are politically or economically constrained, provided that supporting governance and resourcing are treated as essential design requirements. Full article

In China’s stock-based renewal agenda, many old residential communities display pronounced intergenerational overlap, in which grandparental childcare becomes a dominant pattern of outdoor-space use. Against the backdrop of age-structure shifts, population ageing, and persistently low fertility, community-level outdoor-space supply, distributive equity, and environmental adaptability have become key concerns in renewal practice. Yet, practitioners still lack a rankable, low-cost, and implementable evaluation-to-decision workflow. Using Xingshe Community in Dalian, China as an empirical case, this study establishes and tests an integrated “NLP–AHP–GBDT” assessment framework. Guided by policy discourse and planning theory, over 50 semi-structured interviews were processed via NLP-based semantic analysis and keyword mining to derive a two-tier indicator set (criterion and indicator layers). Seven specialists then applied the analytic hierarchy process to elicit indicator weights, and a resident survey was administered to generate weighted performance scores for diagnosing deficiencies. In the feedback-validation stage, we adopted both a qualitative Framework Method and a quantitative GBDT approach, first using the Framework Method to conduct feedback validation based on community residents’ open-ended evaluations. Subsequently, gradient boosting decision trees were used for supervised verification with renewal-scenario data, providing empirical backing for the weighting scheme and the resulting priority order for interventions. The findings suggest that outdoor spaces are broadly serviceable but fall short in intergenerational friendliness, reflecting a structural misalignment between intergenerational activity patterns and spatial provision. Based on the validated priorities, the study proposes modular, incremental micro-renewal measures focusing on safety and emergency accessibility, environmental comfort and caregiving–recreation coupling, and place identity with community organizational mobilization. Full article

This paper defends the thesis that LLM hallucinations are best explained as a truth representation problem: Current models lack an internal representation of propositions as truth-bearers, so truth and falsity cannot constrain generation in the way factual discourse requires. It begins by surveying leading explanations—computational limits on self-verification, deficiencies in training data as truth sources, and architectural factors—and argues that they converge on the same underlying representational deficit. Next, it reconstructs the philosophical background of current LLM design, showing how optimization for fluent continuation aligns with coherence-style evaluation and with broadly structuralist, relational semantics, before turning to David Chalmers’s recent attempt to secure propositional interpretability by drawing on Davidson/Lewis-style radical interpretation and by locating propositional content in “middle-layer” structures; it argues that this approach downplays the ubiquity of hallucination and inherits instability from post-training edits. Finally, the paper offers a positive proposal: Atomic propositions should be represented in the basic vector layer, reviving a logical atomist program as a principled route to reducing hallucination. Full article

Armed conflict management increasingly demands new normative and strategic frameworks that preserve human life while maintaining effective deterrence capabilities. This study develops a multidisciplinary framework for rethinking armed conflict through the concept of just peace, integrating theology, ethics, law, technology, and empirical communication analysis. The research analyzes 7957 YouTube videos from NATO, the United Nations, and the Vatican, published over two years, employing semantic network analysis, modularity-based community detection, and sentiment analysis to identify emerging discourse patterns around peace, technology, and regulatory complexity. The findings suggest that contemporary socio-technological conditions are increasingly framed in ways that open a discursive space for rethinking conflict management beyond exclusive reliance on large-scale lethal force. Positive messaging correlates with higher audience engagement, while concepts such as law, ethics, religion, and technical standards emerge as interconnected regulatory domains. The study concludes that just peace is not naïve pacifism but a strategic, normatively grounded reorientation in contemporary deterrence thinking. Effective implementation requires integrated regulatory frameworks combining legal norms, ethical principles, religious values, and technical standards. The evolving technological landscape may allow deterrence systems to move beyond exclusive reliance on lethal force toward more humane and efficient conflict-management mechanisms. Full article

This paper investigates the diachronic development of the sequence assim que (lit. ‘such that’) in the history of Portuguese, with a comparative perspective on the parallel construction así que in Spanish. A corpus-based approach was employed, analyzing approximately 1800 tokens from the Corpus do Português: Historical Genres, spanning eight centuries of written European Portuguese. The results show that assim que remained highly analyzable until the end of the Old Portuguese period, with the adverb assim often followed by a complement or result clause. The grammaticalization of assim que appears to have evolved partly independently from standalone assim. While Portuguese and Spanish share many uses of the construction, modern European Portuguese has diverged, with assim que losing its status as a discourse marker. This change is best explained by the frequent use of cleft constructions (e.g., foi assim que), which reanalyzed que as a subordinating connector, undoing the earlier single-unit interpretation. These findings suggest that even deeply entrenched grammaticalization processes may undergo retraction when the semantic analyzability of component elements allows it. Full article

The proliferation of generative artificial intelligence (AI) as an autonomous recommendation agent fundamentally challenges traditional paradigms of marketing communication. As AI systems increasingly mediate consumer–brand relationships, understanding how artificial agents construct persuasive discourse—distinct from human communicators—becomes critical for developing effective dual-channel marketing strategies. Grounded in Source Credibility Theory and the Computers Are Social Actors (CASA) paradigm, this study investigates the semantic and structural divergence between AI-generated product recommendations and human influencer marketing messages in social commerce contexts. Employing a mixed-methods computational approach integrating term frequency analysis, TF-IDF weighting, Latent Dirichlet Allocation (LDA) topic modeling, and BERT-based contextualized semantic embedding analysis (KR-SBERT), we examined 330 Instagram influencer posts and 541 AI-generated responses concerning inner beauty enzyme products—a hybrid category combining functional health claims with hedonic beauty appeals—in the Korean social commerce market. AI-generated responses were collected through a systematically designed query protocol with empirically grounded prompts derived from actual consumer search behaviors, and analytical robustness was verified through sensitivity analyses across multiple parameter thresholds. Our findings reveal a fundamental divergence in persuasive architecture: human influencers construct experiential narratives exhibiting message characteristics typically associated with peripheral-route cues (sensory descriptions, emotional testimonials, social context), while AI recommendations employ systematic, evidence-based discourse exhibiting message characteristics typically associated with central-route argumentation (functional mechanisms, ingredient specifications, objective criteria). Topic modeling identified four distinct thematic clusters for each source type: human discourse centers on embodied experience and relational consumption, whereas AI discourse organizes around informational utility and rational decision support. Jensen–Shannon Divergence analysis (JSD = 0.213 bits) confirmed moderate distributional divergence, while chi-square testing (χ² = 847.23, p < 0.001) and Cramér’s V (0.312, indicating a medium-to-large effect) demonstrated statistically significant and substantively meaningful differences. These findings extend CASA theory by demonstrating that AI recommendation agents develop a characteristic “AI communication signature” distinguishable from human persuasion patterns. We propose an integrated Dual-Agent Persuasion Proposition—synthesizing CASA, ELM, and Source Credibility perspectives—suggesting that AI and human recommenders serve complementary functions across different stages of the consumer decision journey—a proposition whose predictions regarding sequential persuasive effectiveness and consumer processing routes await experimental validation. These findings carry implications for AI content strategy optimization, platform design, and emerging regulatory frameworks for AI-generated content labeling. Full article

The digital nomad economy—the ecosystem in which professional skills are traded through online platforms independent of geographic co-location—dynamically recombines skills into project-based portfolios with absent firm-level hierarchy. Yet it remains shaped by platform taxonomies, interfaces, and ranking/recommendation incentives. This study examines the emergent structure within this setting using the Semantic-Structural Systems Analysis (S²SA) framework, which integrates LLM-assisted skill extraction, transformer-based semantic embeddings, and multi-layer network analysis. We analyze a dual-source dataset comprising approximately 50,000 public Upwork profiles from a top-rated/high-earning segment (January–March 2023) and 2.0 million Reddit posts and comments (2018–2023) from remote-work and digital-nomad communities. The resulting skill network exhibits a pronounced core–periphery organization and modular “skill ecotopes” corresponding to coherent functional specializations. In predictive models of skill-level effective hourly rates, semantic brokerage and semantic diversity function as robust predictors of higher rates, significantly outperforming popularity-only baselines. Longitudinal discourse analyses surrounding the COVID-19 pandemic and the generative AI shock reveal rapid attentional shifts followed by the emergence and recombination of new skill clusters. We interpret these results as evidence consistent with constrained self-organization in platform-mediated labor markets. To support replication, prompts, parameters, and robustness checks are fully reported. Full article

Social media has become a key venue where PhD graduates seek career information, compare experiences, and negotiate uncertainty. Drawing on information behavior and sense-making perspectives, this study examines how returnee PhDs from non-core study destinations discuss employment challenges in China’s academic labor market when credential signals are contested. Using Korean-trained PhDs as a theoretically motivated exemplary case, we collected 1149 publicly available posts from Xiaohongshu, a Chinese social media platform, and applied BERTopic to identify latent themes, followed by qualitative close reading of representative posts to interpret discourse functions. The model yielded ten topics, and semantic association analysis indicates substantial overlap among high-frequency topics, suggesting intertwined concerns rather than neatly separated issue domains. The four most prevalent topics account for 72.06% of the corpus, centering on credential recognition, job-search pathways, informal screening rules, and intersecting age- and gender-related pressures. Qualitative readings further reveal recurring discursive moves, including exposing tacit hiring heuristics, contesting stigmatizing labels (e.g., “water PhD,” a derogatory term implying low-quality credentials), and exchanging actionable strategies across regions and career tracks. Overall, the findings point to discursive convergence under evaluation uncertainty: when formal criteria are ambiguous and institutional signals are unreliable, participants turn to social media to stabilize expectations by triangulating cases and iteratively refining shared interpretations of the job market. This study contributes empirical evidence on uncertainty-driven information practices in highly educated labor markets and demonstrates the value of combining topic modeling with qualitative interpretation to capture online collective sense-making. Full article

Why do certain frames gain prominence while others become marginalized in public discourse about emerging technologies? Existing research shows that policy adoption serves as a powerful discursive signal that shapes how issues are interpreted. Yet prior work generally assumes that these framing effects unfold uniformly across jurisdictions. This paper argues instead that the discursive impact of policy adoption is contingent on the interjurisdictional landscape. Integrating insights from policy diffusion theory, I propose that interstate competitive pressure conditions how strongly policy adoption reshapes public discourse. To evaluate this argument, I analyze how autonomous vehicle (AV) policy adoption influences local media framing across U.S. states from 2012 to 2022. Using a dataset of 13,171 news articles, I classify economic, technological, and social/ethical frames with Sentence-BERT, a state-of-the-art semantic model, and estimate causal effects using a staggered difference-in-differences design. The results reveal stark contextual variation: in high-competition states, policy adoption increases economic framing while reducing social and ethical framing, whereas technological framing remains largely unchanged; by contrast, low-competition states exhibit minimal shifts across all frame types. These findings show that the framing effects of policy adoption are relational and context-dependent, advancing research on policy feedback, diffusion, and the politics of emerging technologies. Full article

We propose a semantic, geometric, and statistical framework for quantifying and ordering textual cohesion in long-form discourse. Sentences are embedded into a semantic similarity graph and Ollivier–Ricci curvature is used to extract sentence- and document-level structural profiles, represented as step functions on a normalized rhetorical-time axis. On this functional space we define the Weighted Utopia Index (wUI), a corpus-relative measure of weighted shortfall from an upper-envelope profile under a dominance-type ordering. The rhetorical-time weighting function is learned self-supervised: we generate controlled sentence-order perturbations with known ordinal coherence degradation and estimate the weight parameters via an ordered probit model on a training split. We evaluate ordering recovery on held-out State of the Union speeches using rank correlations, pairwise and adjacent ordering accuracy, and violation-localization diagnostics with bootstrap uncertainty. Across these criteria, wUI systematically outperforms embedding-only adjacent-similarity baselines, while a Nash-type aggregation provides an interpretable semantic–structural trade-off score. An application to later-period speeches illustrates how the method yields interpretable cohesion rankings and curvature-profile diagnostics without requiring external annotations. Full article

This study explores how digital media ecosystems shape collective visions of the future under conditions of rapid technological innovation and the growing influence of artificial intelligence (AI). Drawing on a large corpus of social media content comprising 50,036,592 tokens, the research examines institutional narratives and user-generated responses through a hybrid methodological framework. This framework combines information-wave detection, network analysis, semantic and associative modeling (TextAnalyst 2.32), and interpretation supported by a large language model (GPT-5). The methodological contribution of the study lies in the integration of network-based and semantic algorithms with AI-driven analytical tools for the examination of large-scale textual data. The findings indicate that media discourses about the future operate as key mechanisms through which societies interpret the environmental, social, and economic consequences of technological change. Institutional actors promote multiple future-oriented models that often conflict with one another at both discursive and practical levels. In contrast, user-generated content reflects widespread fear, skepticism, and distrust. Prominent themes include nostalgia for the past, anxiety about socio-economic and environmental consequences, and concerns related to expanding forms of digital control. The analysis also reveals divergent perspectives on urban development. Positive narratives emphasize ecological balance, a comfortable urban environment, thoughtfully designed mixed-use development, and solutions to transportation challenges. Negative narratives, by contrast, focus on over-densification, environmental degradation, and the erosion of privacy in technologically saturated urban spaces. Full article

Multimodal Learning Analytics (MMLA) is an extension of Learning Analytics that combines multiple data streams such as audio, video, physiological signals, logs, and spatial trails to analyze learning processes that cannot be easily captured through any single modality. This review synthesizes research on sensing and instrumentation, feature extraction, multimodal fusion, modeling approaches, and end-to-end systems that provide feedback and support reflection. We also discuss how generative AI and Large Language Models (LLMs) increasingly improve MMLA pipelines by enabling scalable semantic and pragmatic analysis of learner discourse and interaction. In addition, we review robustness issues that arise when working with real-world data (e.g., noise, missing data, and scalability) and responsible deployment issues such as privacy and student-focused views of fairness, accountability, transparency, and ethics (FATE). Full article

Stance detection has emerged as an essential tool in natural language processing for understanding how individuals express agreement, disagreement, or neutrality toward specific targets in social and online discourse. It plays a crucial role in bilingual and multilingual environments, including English-Arabic social media ecosystems, where differences in language structure, discourse style, and data availability pose significant challenges for reliable stance modelling. Existing approaches often struggle with target awareness, cross-lingual generalization, robustness to noisy user-generated text, and the interpretability of model decisions. This study aims to build a reliable, explainable target-aware bilingual stance-detection framework that generalizes across heterogeneous stance formats and languages without retraining on a dataset specific to the target language. Thus, a unified dual-encoder architecture based on mDeBERTa-v3 is proposed. Cross-language contrastive learning offers an auxiliary training objective to align English and Arabic stance representations in a common semantic space. Robustness-oriented regularization is used to mitigate the effects of informal language, vocabulary variation, and adversarial noise. To promote transparency and trustworthiness, the framework incorporates token-level rationale extraction, enables fine-grained interpretability, and supports analysis of hallucination. The proposed model is tested on a combined bilingual test set and two structurally distinct zero-shot benchmarks: MT-CSD and AraStance. Experimental results show consistent performance, with accuracies of 85.0% and 86.8% and F1-scores of 84.7% and 86.8% on the zero-shot benchmarks, confirming stable performance and realistic generalization. Ultimately, these findings reveal that effective bilingual stance detection can be achieved via explicit target conditioning, cross-lingual alignment, and explainability-driven design. Full article