Information

27 pages, 18731 KB

Open AccessArticle

Intelligent Analysis of Data Flows for Real-Time Classification of Traffic Incidents

by Gary Reyes, Roberto Tolozano-Benites, Cristhina Ortega-Jaramillo, Christian Albia-Bazurto, Laura Lanzarini, Waldo Hasperué, Dayron Rumbaut and Julio Barzola-Monteses

Information 2026, 17(3), 310; https://doi.org/10.3390/info17030310 - 23 Mar 2026

Viewed by 530

Abstract

Social media platforms have been established as relevant sources of real-time information for urban traffic analysis. This study proposes an intelligent framework for the classification and spatiotemporal analysis of traffic incidents based on semi-synthetic data streams constructed from historical geolocated seeds for controlled [...] Read more.

Social media platforms have been established as relevant sources of real-time information for urban traffic analysis. This study proposes an intelligent framework for the classification and spatiotemporal analysis of traffic incidents based on semi-synthetic data streams constructed from historical geolocated seeds for controlled validation, utilizing real reports from platforms such as X and Telegram. The approach integrates adaptive machine learning and incremental density-based clustering. An Adaptive Random Forest (ARF) incremental classifier is used to identify the type of incident, allowing for continuous updating of the model in response to changes in traffic flow and concept drift. The classified events are then processed using DenStream, a clustering algorithm that incorporates a temporal decay mechanism designed to identify dynamic spatial patterns and discard older information. The evaluation is performed in a controlled streaming simulation environment that replicates the dynamics of cities such as Panama and Guayaquil. The proposed framework demonstrated robust quantitative performance, achieving a prequential accuracy of up to 86.4% and a weighted F1-score of 0.864 in the Panama scenario, maintaining high stability against semantic noise. The results suggest that this hybrid architecture is a highly viable approach for urban traffic monitoring, providing useful information for Intelligent Transportation Systems (ITS) by processing authentic social signals. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Figure 1

25 pages, 913 KB

Open AccessArticle

Multi-Scale Spatiotemporal Fusion and Steady-State Memory-Driven Load Forecasting for Integrated Energy Systems

by Yong Liang, Lin Bao, Xiaoyan Sun and Junping Tang

Information 2026, 17(3), 309; https://doi.org/10.3390/info17030309 - 23 Mar 2026

Viewed by 449

Abstract

Load forecasting for Integrated Energy Systems (IESs) is critical to enabling multi-energy coordinated optimization and low-carbon scheduling. Facing multi-load types and multi-site high-dimensional heterogeneous data, there remains a global learning challenge stemming from insufficient representation of spatiotemporal coupling features. In response to the [...] Read more.

Load forecasting for Integrated Energy Systems (IESs) is critical to enabling multi-energy coordinated optimization and low-carbon scheduling. Facing multi-load types and multi-site high-dimensional heterogeneous data, there remains a global learning challenge stemming from insufficient representation of spatiotemporal coupling features. In response to the multi-source heterogeneous characteristics of IES loads, this paper designs a Spatiotemporal Topology Encoder that maps load data into a tensorized multi-energy spatiotemporal topological representation via fuzzy classification and multi-scale ranking. In parallel, we construct a MultiScale Hybrid Convolver to extract multi-scale, multi-level global spatiotemporal features of multi-energy load representations. We further develop a Temporal Segmentation Transformer and a Steady-State Exponentially Gated Memory Unit, and design a jointly optimized forecasting model that enforces global dynamic correlations and local, steady-state preservation. Altogether, we propose a multi-scale spatiotemporal fusion and steady-state memory-driven load forecasting method for integrated energy systems (MSTF-SMDN). Extensive experiments on a public real-world dataset from Arizona State University demonstrate the superiority of the proposed approach: compared to the strongest baseline, MSTF-SMDN reduces cooling load RMSE by 16.09%, heating load RMSE by 12.97%, and electric load RMSE by 6.14%, while achieving R² values of 0.99435, 0.98701, and 0.96722, respectively, confirming its feasibility, efficiency, and promising potential for multi-energy load forecasting in IES. Full article

(This article belongs to the Special Issue Emerging Applications of Machine Learning in Healthcare, Industry, and Beyond)

► Show Figures

Figure 1

14 pages, 1332 KB

Open AccessArticle

Leakage-Free Evaluation for Employee Attrition Prediction on Tabular Data

by Ana Maria Căvescu and Alina Nirvana Popescu

Information 2026, 17(3), 308; https://doi.org/10.3390/info17030308 - 23 Mar 2026

Viewed by 453

Abstract

In the context of employee attrition prediction using imbalanced tabular data, we propose a reproducible, leakage-aware evaluation protocol and validate it on the IBM HR Attrition dataset. We perform the train/test split prior to any rebalancing; SMOTE (Synthetic Minority Over-sampling Technique) is applied [...] Read more.

In the context of employee attrition prediction using imbalanced tabular data, we propose a reproducible, leakage-aware evaluation protocol and validate it on the IBM HR Attrition dataset. We perform the train/test split prior to any rebalancing; SMOTE (Synthetic Minority Over-sampling Technique) is applied exclusively within the training portion of each fold in stratified 5-fold cross-validation, while the test set remains untouched. One-Hot Encoding is performed consistently using pd.get_dummies. We benchmark Logistic Regression, Random Forest, ExtraTrees, LightGBM, and XGBoost using imbalance-aware metrics: F1 for the minority class, PR-AUC reported as Average Precision (AP), and ROC-AUC reported both in cross-validation and on the held-out test set. XGBoost attains the best mean AP in cross-validation (0.556 ± 0.056). Logistic Regression achieves the highest mean F1 (0.439 ± 0.048), while LightGBM yields the best mean ROC-AUC (0.791 ± 0.026). On the test set, XGBoost achieves a precision value of 0.65 and a recall value of 0.45 at a fixed threshold of 0.5. Overall, the results highlight a trade-off between stable minority-class detection (Logistic Regression) and stronger risk ranking performance (boosting models) under class imbalance. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Figure 1

25 pages, 1062 KB

Open AccessArticle

The Information Efficiency Metric (IEM): An Info-Metric Approach for Quantifying AI Language Model Performance

by Ljerka Luić, Maja Barbić and Marijana Rončević

Information 2026, 17(3), 307; https://doi.org/10.3390/info17030307 - 22 Mar 2026

Viewed by 647

Abstract

The interaction between humans and artificial intelligence has become a critical channel for information exchange, yet no quantitative, theoretically grounded framework exists for measuring information efficiency in human–AI communication. This study empirically validated an info-metrics framework operationalizing information efficiency through three dimensions—information density [...] Read more.

The interaction between humans and artificial intelligence has become a critical channel for information exchange, yet no quantitative, theoretically grounded framework exists for measuring information efficiency in human–AI communication. This study empirically validated an info-metrics framework operationalizing information efficiency through three dimensions—information density (D), relevance (R), and redundancy (Q)—synthesized into an information efficiency metric (IEM). We analyzed 60 AI responses from ChatGPT 5.2 and Claude Opus 4.5 across factual, analytical, and creative question types using combined coding, automated structural measures, and human evaluation of informational units. The results showed that information density and relevance positively contributed to IEM, while redundancy had a negative contribution. Efficiency varied by task type, with factual prompts showing the highest variability across models and highest efficiency. Contrary to expectations, creative responses did not exhibit higher redundancy, suggesting that expressive diversity does not necessarily constitute informational noise. The framework offers a task-sensitive, theoretically grounded approach to evaluating human–AI information exchange beyond correctness or subjective quality judgment, supporting systems-oriented optimization of conversational AI protocols. Full article

(This article belongs to the Special Issue Multimodal Human-Computer Interaction)

► Show Figures

Figure 1

14 pages, 649 KB

Open AccessArticle

Digitalisation, Digital Governance, and Eco-Innovation: Evidence from Cross-Country Data in 2022

by Keisuke Kokubun

Information 2026, 17(3), 306; https://doi.org/10.3390/info17030306 - 22 Mar 2026

Viewed by 391

Abstract

This study examines the relationship between digitalisation and eco-innovation across countries, with a particular focus on the role of digital government and digital standardisation. Using cross-country data for 2022, eco-innovation is proxied by environment-related patenting activity, while digitalisation is measured using the United [...] Read more.

This study examines the relationship between digitalisation and eco-innovation across countries, with a particular focus on the role of digital government and digital standardisation. Using cross-country data for 2022, eco-innovation is proxied by environment-related patenting activity, while digitalisation is measured using the United Nations E-Government Development Index (EGDI). Employing a combination of ordinary least squares, population-weighted regressions, spline specifications, and quantile regressions, we document three main findings. First, digitalisation is positively and robustly associated with eco-innovation across countries. Second, the relationship is non-linear, with marginal effects that strengthen at higher levels of digital development, suggesting important complementarities between digital capabilities and national innovation systems. Third, the association between digitalisation and eco-innovation is heterogeneous across the distribution of eco-innovation, with particularly strong associations observed among countries with intermediate levels of innovative activity. Taken together, these findings suggest that digitalisation is systematically associated with eco-innovation across countries and indicate the potential relevance of digital governance and digital standardisation to sustainable technological change. Full article

(This article belongs to the Special Issue Standards Digitisation and Digital Standardisation)

► Show Figures

Figure 1

16 pages, 2164 KB

Open AccessArticle

Biometric Identification Under Different Emotions via EEG: A Deep Learning Approach

by Zhyar Abdalla Jamal and Azhin Tahir Sabir

Information 2026, 17(3), 305; https://doi.org/10.3390/info17030305 - 22 Mar 2026

Viewed by 575

Abstract

Electroencephalography (EEG) has attracted growing interest as a biometric modality because it reflects ongoing brain activity and is inherently difficult to counterfeit. At the same time, EEG signals are influenced by internal conditions such as emotions, which may affect identification stability, particularly when [...] Read more.

Electroencephalography (EEG) has attracted growing interest as a biometric modality because it reflects ongoing brain activity and is inherently difficult to counterfeit. At the same time, EEG signals are influenced by internal conditions such as emotions, which may affect identification stability, particularly when recordings are obtained using portable consumer-grade systems. This study examines how emotional states influence EEG-based biometric performance and evaluates deep learning architectures to determine an effective modeling approach for cross-emotion robustness. EEG data were collected from 65 participants using a 14-channel Emotiv EPOC X headset, with 54 subjects retained after self-reported emotional validation. Recordings were acquired under neutral, positive, and negative visual stimuli. To address variability associated with portable acquisition, preprocessing made use of the device’s internal signal quality metrics to select reliable segments, compensate for degraded regions, and reduce noise. Among the evaluated models, a Bidirectional Long Short-Term Memory (BiLSTM) network enhanced with Convolutional Block Attention Module (CBAM) and Multi-Head Self-Attention (MHSA) achieved highest performance in our experiments. The model was trained on neutral-state data and subsequently evaluated under emotional conditions. It reached 95.91% accuracy in the neutral condition and maintained high performance under positive (94.31%) and negative (92.99%) states. Despite a modest decline under negative stimuli, identification performance remained stable. These findings support the feasibility of robust EEG-based biometric authentication using consumer-grade devices in realistic settings. Full article

(This article belongs to the Section Biomedical Information and Health)

► Show Figures

Graphical abstract

22 pages, 1384 KB

Open AccessArticle

Deriving Empirically Grounded NFR Specifications from Practitioner Discourse: A Validated Methodology Applied to Trustworthy APIs in the AI Era

by Apitchaka Singjai

Information 2026, 17(3), 304; https://doi.org/10.3390/info17030304 - 22 Mar 2026

Cited by 1 | Viewed by 489

Abstract

Specifying non-functional requirements (NFRs) for rapidly evolving domains such as trustworthy APIs in the AI era is challenging as best practices emerge through practitioner discourse faster than traditional requirements engineering can capture them. We present a systematic methodology for deriving prioritized NFR specifications [...] Read more.

Specifying non-functional requirements (NFRs) for rapidly evolving domains such as trustworthy APIs in the AI era is challenging as best practices emerge through practitioner discourse faster than traditional requirements engineering can capture them. We present a systematic methodology for deriving prioritized NFR specifications from multimedia practitioner discourse combining AI-assisted transcript analysis, grounded theory principles, and Theme Coverage Score (TCS) validation. Our five-task approach integrates purposive sampling, automated transcription with speaker diarization, grounded theory coding extracting stakeholder-specific themes with TCS quantification, MoSCoW prioritization using empirically derived thresholds (Must Have ≥85%, Should Have 65–84%, Could Have 45–64%, and Won’t Have <45%), and NFR specification consistent with ISO/IEC 25010:2023 principles of stakeholder perspective, measurable quality criteria, and explicit rationale. Applying this methodology to 22 expert presentations on trustworthy APIs yields Weighted Coverage Score of 0.71 and 30 prioritized NFR specifications across five trustworthiness dimensions. MoSCoW classification produces 11 Must Have requirements (Robustness and Transparency), 9 Should Have, 6 Could Have, and 4 Won’t Have. The analysis reveals systematic disparities where Fairness contributes zero Must Have or Should Have requirements due to insufficient practitioner consensus. Each NFR emphasizes stakeholder perspective, measurable quality criteria, and explicit rationale, enabling systematic verification. The validated methodology with complete replication package enables empirically grounded, prioritized NFR derivation from practitioner discourse in any rapidly evolving domain. Full article

(This article belongs to the Special Issue Intelligent Software Engineering: Synergy Between AI and Software Engineering)

► Show Figures

Graphical abstract

29 pages, 722 KB

Open AccessArticle

ChatGPT-Assisted Learning Effectiveness and Academic Achievement: A Mechanism-Based Model in Higher Education

by Ahmed Mohamed Hasanein and Bassam Samir Al-Romeedy

Information 2026, 17(3), 303; https://doi.org/10.3390/info17030303 - 21 Mar 2026

Viewed by 972

Abstract

This study examines the impact of ChatGPT-assisted learning on the academic achievement of hospitality and tourism students in Egyptian public universities, with particular emphasis on the mediating roles of perceived usefulness and self-regulated learning. Drawing conceptually on the Technology Acceptance Model (TAM), the [...] Read more.

This study examines the impact of ChatGPT-assisted learning on the academic achievement of hospitality and tourism students in Egyptian public universities, with particular emphasis on the mediating roles of perceived usefulness and self-regulated learning. Drawing conceptually on the Technology Acceptance Model (TAM), the study adopts a contextualized framework that emphasizes perceived usefulness while incorporating ChatGPT-assisted learning effectiveness as a learning-oriented driver within generative AI-supported educational environments. A quantitative research design was employed using an online survey administered to students who actively used ChatGPT for academic purposes. A total of 689 valid responses were collected from nine public universities and analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) to test the proposed hypotheses. The findings indicate that ChatGPT-Assisted Learning Effectiveness (CALE) has a statistically significant and positive direct effect on academic achievement (AA; β = 0.386, T = 3.946, p < 0.001, 95% CI = 0.192–0.561) and strongly predicts perceived usefulness (β = 0.673, T = 9.274, p < 0.001, 95% CI = 0.581–0.742) and self-regulated learning (β = 0.707, T = 10.734, p < 0.001, 95% CI = 0.621–0.779). In turn, PU (β = 0.281, T = 3.854, p < 0.001, 95% CI = 0.142–0.417) and SRL (β = 0.220, T = 2.418, p = 0.016, 95% CI = 0.041–0.356) significantly enhance academic achievement. Mediation analyses further confirm that PU (β = 0.189, T = 2.366, p = 0.018, 95% CI = 0.031–0.284) and SRL (β = 0.156, T = 3.699, p < 0.001, 95% CI = 0.102–0.301) partially mediate the relationship between CALE and academic achievement. These findings offer important theoretical insights by contextualizing TAM’s performance-related logic within generative AI-driven learning environments and refining its application to academic outcome settings, while highlighting self-regulated learning as a critical explanatory mechanism. From a practical perspective, the study provides valuable implications for educators and policymakers by emphasizing the need to promote students’ perceived usefulness of ChatGPT and foster learner autonomy, positioning generative AI as a powerful pedagogical support tool for enhancing academic success in hospitality and tourism education. Full article

(This article belongs to the Special Issue Trends in Artificial Intelligence-Supported E-Learning)

► Show Figures

Figure 1

40 pages, 15903 KB

Open AccessArticle

A Unified Clustering-Based Anonymization for Privacy-Preserving Data Publishing with Multidimensional Privacy Quantification

by Anselme Herman Eyeleko, Tao Feng and Yan Yan

Information 2026, 17(3), 302; https://doi.org/10.3390/info17030302 - 20 Mar 2026

Viewed by 416

Abstract

As widely adopted privacy models in privacy-preserving data publishing (PPDP), k-anonymity and ℓ-diversity have been extensively studied by researchers to enable the release of useful information while preserving data privacy. However, existing methods suffer from several limitations. They often rely on [...] Read more.

As widely adopted privacy models in privacy-preserving data publishing (PPDP), k-anonymity and ℓ-diversity have been extensively studied by researchers to enable the release of useful information while preserving data privacy. However, existing methods suffer from several limitations. They often rely on single-dimensional privacy models and lack unified metrics for accurately quantifying privacy leakages. Many approaches overlook the impact of semantic similarity and adversarial prior and posterior beliefs among sensitive attributes and frequently employ suboptimal similarity measures that fail to account for the heterogeneous nature of quasi-identifiers, thereby degrading both privacy protection and data utility. To address these challenges, this paper proposes CAMDP, a unified clustering-based anonymization method for privacy-preserving data publishing with multidimensional privacy quantification. CAMDP constructs equivalence classes that satisfy k-anonymity while simultaneously enhancing sensitive attribute diversity, reducing semantic similarity, and limiting divergence between prior and posterior adversarial beliefs. A unified multidimensional metric is introduced to jointly quantify privacy leakage and information loss, guiding the anonymization process. Additionally, a similarity-aware distance metric tailored to mixed-type quasi-identifiers is employed to reduce information loss. Experimental results on three benchmark datasets, Adult, Careplans, and Airline, demonstrate that CAMDP consistently outperforms state-of-the-art methods. Across all tested configurations, CAMDP achieves the lowest average privacy leakage (0.1235, 0.0795, and 0.1855, respectively), lower average information loss (0.626, 0.636, and 0.60, respectively), and the lowest average intra-cluster dissimilarity (0.586, 0.635, and 0.573, respectively), while maintaining competitive execution time across the three datasets. Full article

(This article belongs to the Special Issue Privacy-Preserving Data Analytics and Secure Computation)

► Show Figures

Graphical abstract

25 pages, 2031 KB

Open AccessArticle

A Hybrid Machine Learning Approach for Classifying Indonesian Cybercrime Discourse Using a Localized Threat Taxonomy

by Firman Arifman, Teddy Mantoro and Dini Oktarina Dwi Handayani

Information 2026, 17(3), 301; https://doi.org/10.3390/info17030301 - 20 Mar 2026

Viewed by 607

Abstract

Indonesia’s rapid digital growth has been accompanied by escalating cyber threats, with public discourse on social media emerging as a critical but underutilized source of threat intelligence. This discourse is characterized by informal language and local nuances that render existing international cybercrime taxonomies [...] Read more.

Indonesia’s rapid digital growth has been accompanied by escalating cyber threats, with public discourse on social media emerging as a critical but underutilized source of threat intelligence. This discourse is characterized by informal language and local nuances that render existing international cybercrime taxonomies ineffective, creating a gap in scalable, locally relevant threat analytics. This study introduces the Indonesian Cybercrime Threat Taxonomy (ICTT), a novel five-dimensional framework tailored to Indonesian online environments. An end-to-end OSINT pipeline was developed to collect 2344 samples from X (formerly Twitter) and YouTube, employing weak supervision with 12 high-precision regex patterns to generate training labels. A state-of-the-art IndoBERT model was fine-tuned on this data, and its performance was compared against rule-based and hybrid classification models. On a manually annotated gold-standard dataset of 600 samples, both the IndoBERT and hybrid models achieved 96.8% accuracy, significantly outperforming the rule-based baseline (66.7%). The models demonstrated strong generalization across both social media platforms, and the hybrid approach provided an effective balance of high performance and interpretability. This research demonstrates that informal public discourse can be systematically transformed into structured threat intelligence. The ICTT and the accompanying hybrid classification system provide a scalable, interpretable, and locally relevant foundation for cyber threat analytics in Indonesia, establishing a methodological blueprint for other low-resource language contexts. Full article

(This article belongs to the Special Issue Information Extraction and Language Discourse Processing)

► Show Figures

Figure 1

16 pages, 1943 KB

Open AccessArticle

Exploring Optical Flow Methods for Automated Fall Detection System

by Simeon Karpuzov, Stiliyan Kalitzin, Stefan Tabakov, Dobromir Tsolyov and Georgi Petkov

Information 2026, 17(3), 300; https://doi.org/10.3390/info17030300 - 20 Mar 2026

Viewed by 418

Abstract

Falls pose severe risks to vulnerable populations, particularly the elderly and individuals with adverse neurological conditions, necessitating reliable and non-obstructive detection systems. While previous multi-modal approaches utilizing video and audio have demonstrated strong performance, they face significant limitations regarding sensitivity to environmental noise. [...] Read more.

Falls pose severe risks to vulnerable populations, particularly the elderly and individuals with adverse neurological conditions, necessitating reliable and non-obstructive detection systems. While previous multi-modal approaches utilizing video and audio have demonstrated strong performance, they face significant limitations regarding sensitivity to environmental noise. This paper presents a robust, video-only fall detection framework that eliminates reliance on acoustic data to enhance universality. We conduct a comprehensive comparative analysis of five optical flow (OF) algorithms—Horn–Schunck, Lucas–Kanade (LK), LK-Derivative of Gaussian, Farneback, and the spectral method SOFIA—to determine the range of applicability of each technique for capturing fall dynamics. Beyond detection accuracy, we investigate the computational efficiency of each configuration. This optimized, privacy-centric pipeline offers a scalable solution for continuous monitoring in home and clinical settings, addressing the critical need for immediate intervention following high-impact falls. Full article

► Show Figures

Figure 1

28 pages, 1859 KB

Open AccessReview

Fluency Illusion: A Review on Influence of ChatGPT in Classroom Settings

by Sachin Kumar, Anna Mikayelyan and Olga Vorfolomeyeva

Information 2026, 17(3), 299; https://doi.org/10.3390/info17030299 - 19 Mar 2026

Viewed by 1861

Abstract

The rapid adoption of generative artificial intelligence tools such as ChatGPT in educational settings has generated both enthusiasm and concern regarding their influence on student learning. While several studies report improvements in efficiency, confidence, and perceived understanding, evidence for durable conceptual learning and [...] Read more.

The rapid adoption of generative artificial intelligence tools such as ChatGPT in educational settings has generated both enthusiasm and concern regarding their influence on student learning. While several studies report improvements in efficiency, confidence, and perceived understanding, evidence for durable conceptual learning and knowledge transfer remains mixed. This article examines these tensions through the concept of fluency illusion, a cognitive phenomenon in which information that is easy to process is mistakenly judged as being well understood. Using a narrative conceptual review approach, this study synthesizes findings from 41 publications identified through searches of Google Scholar, Scopus, Web of Science, and ERIC covering the period from 2022 to early 2026. The reviewed literature includes 28 empirical studies, nine conceptual or theoretical analyses, and four review articles addressing the use of ChatGPT in educational contexts. Across domains such as writing and language learning, STEM problem solving, feedback and tutoring, and assessment, the literature shows a recurring pattern in which fluent AI-generated responses increase learners’ confidence without consistently improving deeper conceptual understanding. Drawing on research from cognitive psychology and metacognition, this review proposes an integrative conceptual account of how fluent AI output may shape learners’ judgments of understanding and influence their engagement with learning tasks. The paper concludes by discussing implications for instructional design, assessment practices, and metacognitive scaffolding, and outlines directions for future research aimed at empirically examining the proposed framework and identifying strategies to reduce fluency-driven misjudgments while preserving the potential benefits of generative AI in education. Full article

(This article belongs to the Special Issue Advancing AI Applications in Education and Engineering: A Multidisciplinary Perspective)

► Show Figures

Figure 1

20 pages, 2618 KB

Open AccessArticle

A Deep Hybrid Recommendation Method for Multimodal Information Integrating Content Generated by Large Language Models

by Chao Duan, Wenlong Zhang, Zhongtao Yu, Senyao Li, Xuelian Wan and Qionghao Huang

Information 2026, 17(3), 298; https://doi.org/10.3390/info17030298 - 18 Mar 2026

Viewed by 538

Abstract

Item description information plays a crucial role in helping users understand the basic situation of an item and is also vital auxiliary information in recommendation systems. Traditional methods obtain this data through platform backend data or web scraping techniques, but these data are [...] Read more.

Item description information plays a crucial role in helping users understand the basic situation of an item and is also vital auxiliary information in recommendation systems. Traditional methods obtain this data through platform backend data or web scraping techniques, but these data are often static, relatively fixed, and insufficiently descriptive. In recent years, large language models (LLMs) like generative pre-trained transformer (GPT) have become powerful tools in natural language processing, bringing new hope for LLM-based recommendations. However, does the text information generated by large language models help improve recommendation accuracy? How can the information produced by generative artificial intelligence be integrated with existing multi-source heterogeneous information? In this paper, we propose a novel deep hybrid recommendation method for multimodal information integrating content generated by large language models (DML). We first explore the use of large language models to generate detailed descriptive information about movies. Next, we perform a weighted fusion of the generated text information with existing movie category information and user demographic data, among other multi-source heterogeneous information. Finally, we use the fused information to predict movie ratings. The results indicate that the multimodal information deep hybrid recommendation method, which integrates content generated by large language models, provides substantial evidence of superior performance relative to existing baseline models. Full article

(This article belongs to the Special Issue Generative AI Transformations in Industrial and Societal Applications)

► Show Figures

Figure 1

33 pages, 2332 KB

Open AccessArticle

EvalHack: Answer-Side Prompt Injection for Probing LLM Exam-Grading Panel Stability

by Catalin Anghel, Marian Viorel Craciun, Adina Cocu, Andreea Alexandra Anghel, Antonio Stefan Balau, Adrian Istrate and Aurelian-Dumitrache Anghele

Information 2026, 17(3), 297; https://doi.org/10.3390/info17030297 - 18 Mar 2026

Viewed by 605

Abstract

Large language models are increasingly used as automated graders, yet their reliability under answer-side manipulation and their behavior in multi-model panels remain insufficiently understood. This paper introduces EvalHack, a matrix benchmark in which a fixed committee of four LLMs grades university-level machine learning [...] Read more.

Large language models are increasingly used as automated graders, yet their reliability under answer-side manipulation and their behavior in multi-model panels remain insufficiently understood. This paper introduces EvalHack, a matrix benchmark in which a fixed committee of four LLMs grades university-level machine learning exam answers under a strict integer-only contract (0–10) grounded in instructor-authored rubric artifacts. The dataset comprises 100 students answering 10 short, open-ended items (1000 answers). For each answer, the evaluation includes a clean version and two content-preserving adversarial variants that operate only on the student text: A₁, a visible coercive suffix appended to the answer, and A₂, a stealth variant that uses Unicode control characters (e.g., zero-width and bidirectional marks) to embed an instruction. EvalHack instruments the full grading pipeline, recording item-level member scores, the committee aggregate, within-panel disagreement, and discrepancies to human grades. Empirically, answer-side edits induce systematic score inflation and stronger top-end concentration, with edited answers clustering near the upper end of the scale. Within-panel disagreement, measured as the range between the highest and lowest member score, varies across conditions, with median Consistency Spread values of 3.0 (clean), 2.0 (A₁), and 6.0 (A₂). Compared to human graders, the panel is more lenient on average (MAE = 1.897; bias human − panel = −1.345). Finally, grouping items by disagreement shows that low-disagreement items exhibit smaller human-panel errors, indicating that within-panel spread can serve as a practical uncertainty signal for routing difficult answers to human review or to larger/more specialized panels. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Graphical abstract

16 pages, 2202 KB

Open AccessArticle

A Hybrid Ensemble Machine Learning Framework with Membership-Function Feature Engineering for Non-Invasive Prediction of HER2 Status in Breast Cancer

by Hassan Salarabadi, Dariush Salimi, Seyed Sahand Mohammadi Ziabari and Mozaffar Aznab

Information 2026, 17(3), 296; https://doi.org/10.3390/info17030296 - 18 Mar 2026

Viewed by 388

Abstract

Accurate determination of human epidermal growth factor receptor 2 (HER2) status is a critical component of breast cancer prognosis and treatment planning. Conventional diagnostic techniques, such as immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH), are clinically established but remain invasive, time-consuming, costly, [...] Read more.

Accurate determination of human epidermal growth factor receptor 2 (HER2) status is a critical component of breast cancer prognosis and treatment planning. Conventional diagnostic techniques, such as immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH), are clinically established but remain invasive, time-consuming, costly, and sensitive to pre-analytical and interpretative variability. Motivated by the need for scalable and data-driven decision-support tools, this study proposes a hybrid ensemble machine learning framework for non-invasive HER2 status prediction using routinely available clinical and immunohistochemical features. A retrospective dataset comprising 624 breast cancer patients from Mahdieh Clinic (Kermanshah, Iran) was analyzed using a structured preprocessing pipeline including normalization and class balancing. The proposed framework integrates multiple tree-based classifiers, Random Forest, XGBoost, and LightGBM, through ensemble strategies and enhances predictive robustness using membership-function feature engineering to capture gradual transitions in clinically relevant biomarkers. Decision threshold optimization was further applied to improve classification balance in borderline cases. The proposed ensemble framework achieved an accuracy of 0.816, an F1-score of 0.814, and an area under the receiver operating characteristic curve (AUC) of 0.862 on a held-out test set, demonstrating performance comparable to the best-performing individual classifier. These results indicate that ensemble learning combined with smooth membership-based feature representations can provide a reliable decision-support framework for HER2 status prediction, although further external validation is required before clinical use. Full article

(This article belongs to the Special Issue Information Management and Decision-Making)

► Show Figures

Figure 1

31 pages, 580 KB

Open AccessArticle

Seeing the Message but Not the Machine: Digital Skepticism and AI Discernment in Online Information Environments

by Lersak Phothong, Anupong Sukprasert, Nattakarn Shutimarrungson and Mehtabhorn Obthong

Information 2026, 17(3), 295; https://doi.org/10.3390/info17030295 - 18 Mar 2026

Viewed by 762

Abstract

Artificial intelligence (AI) increasingly mediates how information is generated, ranked, and circulated in digital environments. However, it remains unclear under what conditions users explicitly articulate recognition of AI involvement in routine news-related discourse. This study examines how digital skepticism and AI-related discernment are [...] Read more.

Artificial intelligence (AI) increasingly mediates how information is generated, ranked, and circulated in digital environments. However, it remains unclear under what conditions users explicitly articulate recognition of AI involvement in routine news-related discourse. This study examines how digital skepticism and AI-related discernment are expressed in naturally occurring social media discussions. Using an exploratory observational design, 6065 user-generated comments from 305 news-related Reddit threads were analyzed through a rule-based framework distinguishing general skepticism, structural suspicion, and explicit AI-related discernment. Within the sampled corpus, generalized digital skepticism is proportionally more visible than explicit attribution to AI-generated or synthetically produced content. Explicit AI-related attribution is unevenly distributed across discourse contexts, appearing more frequently in technology-oriented communities and remaining limited in mainstream news-related discussions. Differences across score-based visibility contexts do not correspond to a consistently higher representation of explicit AI attribution. These findings indicate a distributional difference between generalized skepticism and publicly articulated recognition of AI mediation. Rather than measuring levels of awareness, the results illuminate the contextual and linguistic conditions under which AI involvement becomes explicitly named in public interaction. By focusing on observable discourse rather than self-reported attitudes, the study provides a corpus-bound account of when AI mediation becomes discursively articulated in algorithmically mediated environments. Full article

► Show Figures

Figure 1

27 pages, 8038 KB

Open AccessArticle

Adaptive Measurement Noise Covariance Estimation for GNSS/INS Tightly Coupled Integration Using a Linear-Attention Transformer with Residual Sparse Denoising and Channel Attentions

by Ning Wang and Fanming Liu

Information 2026, 17(3), 294; https://doi.org/10.3390/info17030294 - 17 Mar 2026

Viewed by 427

Abstract

Tightly coupled GNSS/INS is a widely adopted architecture for UAVs and ground vehicles. In this study, a Kalman-filter-based fusion framework integrates inertial data with satellite observables, including pseudorange and Doppler-derived range rate, to sustain precise navigation when GNSS quality degrades. A key bottleneck [...] Read more.

Tightly coupled GNSS/INS is a widely adopted architecture for UAVs and ground vehicles. In this study, a Kalman-filter-based fusion framework integrates inertial data with satellite observables, including pseudorange and Doppler-derived range rate, to sustain precise navigation when GNSS quality degrades. A key bottleneck is that many pipelines rely on fixed or overly simplified measurement-noise covariance models, which cannot track the nonstationary statistics of real observations. To address this issue, we develop an adaptive covariance estimator built on a Transformer enhanced with three modules: a Linear-Attention layer, a Residual Sparse Denoising Autoencoder (R-SDAE), and a lightweight residual channel-attention block (LRCAM). The estimator predicts the measurement-noise covariance online. R-SDAE distills sparse, outlier-resistant features from noisy ephemeris; LRCAM reweights informative channels via residual gating; and Linear Attention preserves long-range spatiotemporal dependencies while reducing attention cost from O(N²) to O(N). A predictive factor further modulates the covariance for improved efficiency and adaptability. Experimental results on real road-test data show that the proposed method achieves sub-meter positioning accuracy in open-sky conditions and preserves meter-level accuracy with improved robustness under GNSS-degraded urban scenarios, outperforming the compared adaptive-filtering baselines and neural covariance estimators and thereby demonstrating superior positioning accuracy and stability. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Figure 1

28 pages, 1092 KB

Open AccessArticle

A Secure and Robust ML Framework for Sequence Classification and Adversarial Evaluation in a Bilateral Carpal Tunnel Syndrome Crossover Dataset

by Pratik Pandurang Kharat, Sufian Al Majmaie, Ghazal Ghajari, Fathi Amsaad and Mohamed I. Ibrahem

Information 2026, 17(3), 293; https://doi.org/10.3390/info17030293 - 17 Mar 2026

Cited by 1 | Viewed by 469

Abstract

Bilateral idiopathic carpal tunnel syndrome (CTS) is a neuromuscular condition involving the compression of the median nerve at both wrists, leading to pain, neurological symptoms, and loss of function. This paper proposes a robust machine-learning framework for a randomized crossover clinical trial comparing [...] Read more.

Bilateral idiopathic carpal tunnel syndrome (CTS) is a neuromuscular condition involving the compression of the median nerve at both wrists, leading to pain, neurological symptoms, and loss of function. This paper proposes a robust machine-learning framework for a randomized crossover clinical trial comparing two physiotherapeutic treatment regimens: stretching followed by myofascial mobilization (S/M) and the reverse sequence (M/S). Instead of making inferences about the superiority of one treatment over another, the treatment regimen serves as a structured analytical label for investigating predictive separability, feature representation, and model stability within a controlled experimental setting. The clinical dataset of 73 patients underwent rigorous preprocessing, including strength feature aggregation and principal component analysis (PCA). Various classifiers were evaluated, with CatBoost achieving an ROC-AUC of 0.985 and a test accuracy of 96.5%, while Random Forest demonstrated strong adversarial robustness with an adversarial accuracy of 96.83%. To assess robustness, clinically constrained perturbations were introduced into the PCA feature space, simulating realistic input variability. The findings indicate that ensemble learning algorithms can capture structured patterns in crossover clinical datasets and remain stable under low-magnitude adversarial perturbations. The study underscores the importance of robustness evaluation and interpretability when applying machine learning models to biomedical data, particularly in small and well-structured clinical cohorts. Full article

(This article belongs to the Section Biomedical Information and Health)

► Show Figures

Graphical abstract

31 pages, 1934 KB

Open AccessReview

Artificial Intelligence for Detecting Electoral Disinformation on Social Media: Models, Datasets, and Evaluation

by Félix Díaz, Nhell Cerna, Rafael Liza and Bryan Motta

Information 2026, 17(3), 292; https://doi.org/10.3390/info17030292 - 17 Mar 2026

Viewed by 1086

Abstract

During elections, information manipulation on social media has accelerated the use of artificial intelligence, yet the evidence is difficult to interpret without an integrated view of methods, data, and evaluation. We mapped 557 English-language journal articles from Scopus and Web of Science, combining [...] Read more.

During elections, information manipulation on social media has accelerated the use of artificial intelligence, yet the evidence is difficult to interpret without an integrated view of methods, data, and evaluation. We mapped 557 English-language journal articles from Scopus and Web of Science, combining performance indicators, science mapping, and a focused full-text synthesis of highly cited papers. The literature grows sharply after 2019, peaks in 2025, and shows geographically uneven production, with collaboration structured around a small set of hubs. The thematic structure suggests that, during the pandemic era, infodemic-related research served as a catalyst, intensifying scientific attention to fake news and disinformation and expanding the associated detection and monitoring agendas. In addition, socio-political harm constructs such as hate speech, extremism, and polarization appear as recurrent and structurally central targets, highlighting that election-relevant work often extends beyond veracity assessment toward monitoring discourse risks. Blockchain also emerges as a novel and adjacent integrity theme, aligned with authenticity and provenance-oriented mitigation rather than mainstream detection pipelines. AI for electoral disinformation is not reducible to veracity classification, as influential studies also target automation and coordinated behavior, verification support, diffusion analysis, and estimation frameworks that focus on exposure and impact. Evaluation remains heterogeneous and is often shaped by benchmark settings, making high accuracy values hard to compare and potentially misleading when labeling quality, topic leakage, or context shift are not characterized. Overall, the findings motivate evaluation protocols that align operational objectives with modeling roles and explicitly address robustness to temporal and platform changes, asymmetric error costs during election windows, and representativeness across electoral contexts and languages, while also guiding future work on emerging integrity challenges and governance-relevant deployment settings. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Figure 1

19 pages, 2255 KB

Open AccessArticle

Empirical Validation of Software Engineering Deadpoints: An Expert Practitioner Survey

by Abdullah A. H. Alzahrani

Information 2026, 17(3), 291; https://doi.org/10.3390/info17030291 - 17 Mar 2026

Viewed by 578

Abstract

A state of terminal stagnation is often reached by software projects despite the presence of advanced tools, and these occurrences are defined within this study as software engineering deadpoints, where the cost of system recovery is frequently found to be higher than the [...] Read more.

A state of terminal stagnation is often reached by software projects despite the presence of advanced tools, and these occurrences are defined within this study as software engineering deadpoints, where the cost of system recovery is frequently found to be higher than the actual value of the software. While many factors are seen to lead toward project failure, it is suggested by the evidence that technical debts are the main cause of such failures. A significant number (23.5%) of these fatal issues is created during the early architectural phases of development, and it is noted that these problems often remain hidden until they become unrecoverable. The data collected during this research show that projects facing technical obstacles (Recovery Score: 4.24) are much harder to save than those suffering with process obstacles (Recovery Score: 5.38). It was also observed that a steady reluctance to refactor old logic and an excessive number of code revisions are seen as the most reliable signs that a project is approaching a point of no return. Because these warning signs are often overlooked by management, the eventual failure of the system is often viewed as an unexpected event rather than a predictable outcome of poor early choices. By defining these terminal states, this work provides those in leadership roles with a method to differentiate between minor delays and total failure, thereby assisting teams in avoiding the heavy economic losses associated with unproductive development paths. Full article

► Show Figures

Figure 1

30 pages, 2796 KB

Open AccessArticle

Information Recovery Under Partial Observation: A Methodological Analysis of Multi-Informant Questionnaire Data

by Nawaphol Thepnarin and Adisorn Leelasantitham

Information 2026, 17(3), 290; https://doi.org/10.3390/info17030290 - 15 Mar 2026

Viewed by 515

Abstract

This study examines information recovery under structured partial observation in multi-informant questionnaire systems. Rather than predicting an external ground truth, we evaluate the recoverability of an operational full-information decision rulewhen only partial informant views are available. In the empirical SNAP-IV calibration study, this [...] Read more.

This study examines information recovery under structured partial observation in multi-informant questionnaire systems. Rather than predicting an external ground truth, we evaluate the recoverability of an operational full-information decision rulewhen only partial informant views are available. In the empirical SNAP-IV calibration study, this reference is intentionally defined as a deterministic function of the combined informant views, so the combined-view result is treated only as an oracle-style ceiling and the substantive analysis concerns how single-view recovery degrades when one informant is withheld. To examine whether a similar qualitative pattern extends beyond this calibration setting, we additionally evaluate a latent-state simulation in which the reference decision is generated from an unobserved latent state and informant views are noisy observations. Across both settings, single-view recoverability declines as inter-rater disagreement increases, whereas combined-view representations remain more stable. In the empirical study, combined-view models achieved near-ceiling recovery performance (e.g., 90.9% for Logistic Regression and 91.3% for MLP), while Teacher-only recovery dropped from approximately 78% to 63% under higher disagreement (

p = 0.0005

, Cohen’s

d = 1.9

). Additional non-learned single-rater score-threshold baselines exhibited the same qualitative degradation pattern, indicating that the effect is not specific to fitted machine learning models. Importantly, this work is methodological: it does not propose new learning algorithms or clinical prediction models, but instead presents a conceptual–methodological framework, together with model-agnostic recoverability quantities, for quantifying missing-view information loss under incomplete, heterogeneous observations. Full article

(This article belongs to the Section Information Theory and Methodology)

► Show Figures

Figure 1

20 pages, 3980 KB

Open AccessArticle

Influence of Input Data Uncertainty on Cellular Automata-Based Wildfire Spread Simulation

by Ioannis Karakonstantis and George Xylomenos

Information 2026, 17(3), 289; https://doi.org/10.3390/info17030289 - 15 Mar 2026

Viewed by 385

Abstract

Cellular automata-based wildfire simulation models are widely used to support fire management, risk assessment, and operational decision-making, due to their efficiency and computational advantages. However, the accuracy of these models heavily depends on the quality of input data provided by the user, including [...] Read more.

Cellular automata-based wildfire simulation models are widely used to support fire management, risk assessment, and operational decision-making, due to their efficiency and computational advantages. However, the accuracy of these models heavily depends on the quality of input data provided by the user, including the composition and geospatial extend of forest fuels, current meteorological conditions and terrain information. This publication examines how quantitative and spatial input data uncertainties affect the estimates of the impacted areas. Using a series of simulation experiments, inaccurate data are introduced to specific input variables (such as the vegetation type and the fuel moisture content) to reflect realistic levels of uncertainty commonly observed in operational scenarios, where users with different cognitive backgrounds fail to properly identify key characteristics of a fire. Model outputs are then compared using spatial and temporal performance metrics, including the rate of spread and burned area extent. The results demonstrate that uncertainties in fuel models and meteorological inputs exert a dominant influence on simulated fire behavior. Our findings highlight the sensitivity of wildfire simulations to compounded input uncertainties and stress the need for improved in-field data acquisition strategies. Full article

(This article belongs to the Section Information Applications)

► Show Figures

Figure 1

22 pages, 3691 KB

Open AccessArticle

Interpreting Interaction Patterns and Cognitive Strategies in LLM-Supported Exploratory Learning: A Mixed-Methods Analysis Using the DOK Framework

by Yiming Taclis Luo, Ting Liu, Patrick Pang, Dana McKay, Shanton Chang and George Buchanan

Information 2026, 17(3), 288; https://doi.org/10.3390/info17030288 - 14 Mar 2026

Viewed by 1560

Abstract

As exploratory learning (EL) is increasingly observed with the use of large language models (LLMs), students demonstrate notably varied levels of engagement and effectiveness when they interact with such LLM-supported learning environments. However, the underlying mechanisms driving these disparities, particularly in how students [...] Read more.

As exploratory learning (EL) is increasingly observed with the use of large language models (LLMs), students demonstrate notably varied levels of engagement and effectiveness when they interact with such LLM-supported learning environments. However, the underlying mechanisms driving these disparities, particularly in how students interact with LLMs, remain underexplored. To address this gap, this observational comparative study systematically investigates the EL strategies of 46 students in two different regions of Asia, classifying 25 distinct strategies across cognitive stages using the Depth of Knowledge model. Our analysis compares strategy usage between high and low-performing student subgroups. The findings reveal: (1) A declining trend in the utilization of EL strategies across ascending cognitive stages. (2) High AWP students employed EL strategies more frequently than their peers, with ten EL strategies exhibiting significant between-group differences. (3) Among students with different AI experience, only a few EL strategies usage and cognitive stages showed significant differences. These insights can help educators and LLM interface designers develop targeted exploratory learning assistance for different types of students and help them build high-level metacognitive processes for effective human–computer interaction. Full article

(This article belongs to the Special Issue Human–Computer Interactions and Computer-Assisted Education)

► Show Figures

Figure 1

20 pages, 1426 KB

Open AccessReview

Profiling Decision-Making Styles Under Healthcare Resource Scarcity: An Interdisciplinary Clustering Approach

by Micaela Pinho, Fátima Leal and Isabel Miguel

Information 2026, 17(3), 287; https://doi.org/10.3390/info17030287 - 14 Mar 2026

Cited by 1 | Viewed by 623

Abstract

Scarcity of healthcare resources requires prioritisation decisions that raise complex ethical, economic, and social challenges. While normative frameworks provide guidance on how such decisions ought to be made, growing evidence suggests that individuals differ substantially in how they approach morally charged allocation choices. [...] Read more.

Scarcity of healthcare resources requires prioritisation decisions that raise complex ethical, economic, and social challenges. While normative frameworks provide guidance on how such decisions ought to be made, growing evidence suggests that individuals differ substantially in how they approach morally charged allocation choices. This study investigates heterogeneity in decision-making styles and support for healthcare prioritisation criteria using an interdisciplinary approach that integrates health economics, social psychology, and computational methods to identify latent decision-making profiles among a sample of adults residing in Portugal. Data were collected from adults residing in Portugal using a structured online questionnaire comprising socio-demographic characteristics, decision-making styles, and preferences elicited through twenty hypothetical healthcare rationing scenarios. The results reveal three meaningful decision-making profiles characterised by different combinations of cognitive styles and ethical prioritisation patterns: analytically oriented decision-makers prioritising health gains; intuitive, context-sensitive decision-makers balancing clinical and social criteria; heuristic-driven decision-makers relying on simpler or less differentiated heuristics. These findings demonstrate that, within this sample, healthcare prioritisation preferences are shaped by systematic variations in decision style rather than a single moral or rational framework. By linking behavioural heterogeneity with ethical decision-making, this study contributes to theoretical debates on healthcare rationing and demonstrates the value of clustering techniques for uncovering latent structures in complex decision data. The results provide insights relevant for the design of decision-support systems and rationing policies, which may be adapted to accommodate heterogeneous decision styles in comparable settings. Full article

(This article belongs to the Topic Machine Learning and Data Mining: Theory and Applications)

► Show Figures

Figure 1

38 pages, 1411 KB

Open AccessArticle

Cybersecurity Digital Twins for Industrial Systems: From Literature Synthesis to Framework Design

by Konstantinos E. Kampourakis, Vasileios Gkioulos and Sokratis Katsikas

Information 2026, 17(3), 286; https://doi.org/10.3390/info17030286 - 12 Mar 2026

Viewed by 1321

Abstract

Digital Twins (DTs) are increasingly recognized as a strategic technology for enhancing cybersecurity in industrial environments, particularly in the face of rising threats targeting Operational Technology (OT). After comparatively examining closely related DT–cybersecurity frameworks to position the contribution within the existing research landscape, [...] Read more.

Digital Twins (DTs) are increasingly recognized as a strategic technology for enhancing cybersecurity in industrial environments, particularly in the face of rising threats targeting Operational Technology (OT). After comparatively examining closely related DT–cybersecurity frameworks to position the contribution within the existing research landscape, this paper presents a systematic literature review and comparative analysis of 19 recent DT-based cybersecurity studies, focusing on their relevance to incident detection and response in sectors such as Industrial Internet of Things (IIoT), manufacturing, and energy. The analysis evaluates each study across multiple dimensions, including attack types, detection and response mechanisms, DT integration, and technology stacks. From this review, we derive a consolidated set of requirements, categorized as functional, non-functional, security-specific, and domain-specific. These requirements serve as the foundation for a novel, cybersecurity-focused, ISO 23247-based framework. The proposed architecture formalizes a DT-enabled incident detection and response lifecycle aligned with ISO 23247. It is explicitly mapped to the derived requirements and detailed with practical implementation considerations. This work contributes a structured, evidence-based approach to DT-based security engineering and offers a reference design for researchers and practitioners aiming to build resilient, adaptive cybersecurity solutions in industrial settings. Full article

(This article belongs to the Special Issue Security and Privacy Approaches Against Cyber Threats: Innovations, Challenges, and Practical Solutions)

► Show Figures

Figure 1

34 pages, 1485 KB

Open AccessArticle

Cognitive Digital Twin Generations: From Foundational Instruments to Meta-Cognitive Ecosystems

by Igor Kabashkin

Information 2026, 17(3), 285; https://doi.org/10.3390/info17030285 - 12 Mar 2026

Viewed by 1046

Abstract

The growing complexity of cyber-physical and socio-technical systems calls for digital twin architectures capable of modeling cognition-driven processes such as perception, reasoning, learning, and reflection. This paper proposes an instrumental and generational framework of cognitive digital twins (CDTs) that formalizes cognition as an [...] Read more.

The growing complexity of cyber-physical and socio-technical systems calls for digital twin architectures capable of modeling cognition-driven processes such as perception, reasoning, learning, and reflection. This paper proposes an instrumental and generational framework of cognitive digital twins (CDTs) that formalizes cognition as an explicit and evolvable system property. The framework defines a stable set of cognitive modeling instruments—cognitive analyzer, cognitive emulator and cognitive orchestrator—and introduces four CDT generations: foundational CDTs, self-adaptive CDTs, collective CDTs and meta-cognitive digital ecosystems. The study focuses on foundational cognition modeling as the primary generation and develops a mathematical framework based on the cognitive maturity index and the ontology consistency index to quantify cognitive behavior and semantic coherence. Convergence analysis and representative application scenarios validate the stability of the proposed model. Higher CDT generations are introduced to establish an evolutionary roadmap toward adaptive, collective, and meta-cognitive digital twins. The proposed framework integrates conceptual taxonomy, instrumental typology, and a methodological roadmap for instrument selection and evolution, providing a unified foundation for modeling cognition-driven systems and extending traditional digital twin paradigms. Full article

(This article belongs to the Special Issue Information-Driven Synergies in the Metaverse and IoT Ecosystems)

► Show Figures

Graphical abstract

15 pages, 1269 KB

Open AccessArticle

Deploying Efficient LLM Agents on Maritime Autonomous Surface Ships: Fine-Tuning, RAG, and Function Calling in a Mid-Size Model

by Yiling Ren, Mozi Chen, Junjie Weng, Shengkai Zhang, Xuedou Xiao and Kezhong Liu

Information 2026, 17(3), 284; https://doi.org/10.3390/info17030284 - 12 Mar 2026

Cited by 1 | Viewed by 795

Abstract

Deploying Large Language Models (LLMs) on Maritime Autonomous Surface Ships (MASS) entails a critical trade-off between reasoning depth, inference latency, and hardware constraints. To fill the existing gap, we introduce MARTIAN (Maritime Agent for Real-time Tactical Inference [...] Read more.

Deploying Large Language Models (LLMs) on Maritime Autonomous Surface Ships (MASS) entails a critical trade-off between reasoning depth, inference latency, and hardware constraints. To fill the existing gap, we introduce MARTIAN (Maritime Agent for Real-time Tactical Inference And Navigation), a 14B-parameter decision support agent engineered for edge deployment on standard vessel hardware (e.g., the NVIDIA Jetson AGX Orin). Central to our approach is the Cognitive Core architecture, which utilizes a verified dataset of 21,800 Chain-of-Thought (CoT) instruction–response pairs to align general linguistic capabilities with maritime procedural logic. Empirical evaluations demonstrate that MARTIAN achieves an overall accuracy of 73.23% (SFT only) and 81.16% (SFT + RAG) on the Bilingual Maritime Multiple-Choice Questionnaire (BM-MCQ), a standardized assessment dataset constructed based on Officer of the Watch (OOW) competencies. Notably, the SFT-only configuration attains 78.53% on pure-logic-intensive COLREG tasks—surpassing the 72B-parameter Qwen-2.5 foundation model in this domain—while maintaining a real-time inference latency of 22.4 ms/token. Crucially, our ablation studies support a nuanced Interference Hypothesis: while RAG significantly enhances factual recall in knowledge-intensive domains (boosting total accuracy from 73.23% to 81.16%), it concurrently introduces semantic noise that degrades performance in pure logic reasoning tasks (e.g., COLREG maneuvering accuracy decreases from 78.53% to 77.36%). On the basis of this finding, we identify and empirically motivate a decoupled cognitive design principle that separates procedural reflexes (via SFT) from declarative knowledge (via RAG). While the full implementation of an adaptive routing mechanism is deferred to future work, the ablation results presented herein offer a validated, cost-effective reference architecture for deploying transparent and regulation-compliant AI on resource-constrained merchant vessels. Full article

(This article belongs to the Special Issue Artificial Intelligence in Autonomous Navigation and Intelligent Unmanned Systems)

► Show Figures

Figure 1

30 pages, 1238 KB

Open AccessArticle

Activation-Guided Layer Selection for LoRA

by Aditya Dawadikar, Pooja Shyamsundar, Rashmi Vishwanath Bhat and Navrati Saxena

Information 2026, 17(3), 283; https://doi.org/10.3390/info17030283 - 12 Mar 2026

Viewed by 1240

Abstract

Low-Rank Adaptation (LoRA) has become a widely adopted parameter-efficient fine-tuning (PEFT) technique for large language models (LLMs). LoRA’s benefits stem from its light weight and modular adapters. Standard LoRA applies adapters uniformly across all Transformer layers, implicitly assuming that each layer contributes equally [...] Read more.

Low-Rank Adaptation (LoRA) has become a widely adopted parameter-efficient fine-tuning (PEFT) technique for large language models (LLMs). LoRA’s benefits stem from its light weight and modular adapters. Standard LoRA applies adapters uniformly across all Transformer layers, implicitly assuming that each layer contributes equally to task adaptation. However, LLMs are found to have internal substructures that contribute in a disproportionate manner. In this work, we provide a theoretical analysis of how LoRA weight updates are influenced by a layer’s activation magnitude. We propose Act-LoRA, a simple activation-guided layer selection strategy for selective Low-Rank Adaptation. We evaluate this strategy for both encoder-only and decoder-only architectures using the GLUE benchmark. Our method achieved a 20% GPUh saving with a 1% drop in GLUE score using DeBERTaV3-Base on a single-instance GPU with 50% less LoRA parameters. It also achieved 2% GPUh savings with a less than 0.15% drop in GLUE score with the Llama-3.1-8B model in Distributed Data Parallel mode with 25% fewer LoRA parameters. Our experiments and analysis show that the compute and memory requirements of LoRA adapters increase linearly with the number of selected layers. We further compare activation-guided selection against gradient-guided importance metrics and show that activation norms yield more stable and reproducible layer rankings across seeds and datasets. Overall, our results demonstrate that activation-guided layer selection is a practical and effective way to improve the efficiency of LoRA fine-tuning, making it immediately compatible with some existing PEFT techniques and distributed training frameworks. Full article

(This article belongs to the Special Issue Modeling in the Era of Generative AI)

► Show Figures

Graphical abstract

37 pages, 6747 KB

Open AccessSystematic Review

AI-Supported Gamification in E-Learning: A Systematic Review of Adaptive Architectures and Cognitive Outcomes

by Aray Kassenkhan, Vassiliy Serbin, Roza Beisembekova, Aigerim Abshukirova and Bayan Mendekina

Information 2026, 17(3), 282; https://doi.org/10.3390/info17030282 - 12 Mar 2026

Cited by 2 | Viewed by 2037

Abstract

The rapid expansion of artificial intelligence (AI) in digital education has transformed gamification from a motivational strategy into a data-driven, adaptive learning paradigm. This systematic review conceptualizes AI-supported gamification as an information-centered ecosystem integrating learning analytics, behavioral modeling, adaptive algorithms, and intelligent feedback [...] Read more.

The rapid expansion of artificial intelligence (AI) in digital education has transformed gamification from a motivational strategy into a data-driven, adaptive learning paradigm. This systematic review conceptualizes AI-supported gamification as an information-centered ecosystem integrating learning analytics, behavioral modeling, adaptive algorithms, and intelligent feedback mechanisms to enhance cognitive development and critical thinking. Following PRISMA 2020 guidelines, a systematic search was conducted across Scopus, Web of Science, ScienceDirect, Google Scholar, and ResearchGate. Peer-reviewed empirical studies published between 2020 and 2025 were considered. Studies were included if they examined gamification in educational contexts with AI-driven or adaptive system components, while non-educational contexts, duplicates, and non-English publications were excluded. After screening and eligibility assessment, 100 studies were included in the final synthesis. The review examines how AI-driven personalization, neurotechnology, predictive modeling, and generative systems reshape the design and effectiveness of gamified e-learning environments. Architectural patterns identified include recommender systems, real-time behavioral adaptation, affect-aware feedback loops, and algorithmic content generation. Across the reviewed studies, AI-supported gamified systems were frequently associated with increased engagement and moderate improvements in executive functions, higher-order reasoning, and adaptive learning pathways. However, challenges related to system transparency, data governance, algorithmic bias, cognitive load management, and equitable access remain significant. The review was not registered. By framing gamification as an adaptive information system rather than solely a pedagogical intervention, this study proposes a structured taxonomy of AI-driven gamified architectures—including data acquisition, user modeling, predictive analytics, and adaptive feedback layers—and outlines research priorities for scalable, ethically grounded, and data-informed e-learning ecosystems. Full article

(This article belongs to the Special Issue Trends in Artificial Intelligence-Supported E-Learning)

► Show Figures

Graphical abstract

55 pages, 17048 KB

Open AccessReview

The Evolution of Visualization Technologies in Healthcare: A Bibliometric Analysis of Studies Published from 1994 to 2025

by Fangzhong Cheng, Chun Yang and Rong Deng

Information 2026, 17(3), 281; https://doi.org/10.3390/info17030281 - 11 Mar 2026

Viewed by 1024

Abstract

Healthcare visualization has become a crucial approach for interpreting complex medical data, supporting informed clinical decision-making, and enhancing public health management. However, existing reviews tend to focus on specific technologies or application scenarios, offering limited insight into the field’s overall knowledge structure, developmental [...] Read more.

Healthcare visualization has become a crucial approach for interpreting complex medical data, supporting informed clinical decision-making, and enhancing public health management. However, existing reviews tend to focus on specific technologies or application scenarios, offering limited insight into the field’s overall knowledge structure, developmental trajectory, and interdisciplinary integration. To address this gap, this study systematically reviews 1121 publications from 1994 to 2025 indexed in the Web of Science Core Collection. By combining bibliometric analysis with qualitative assessment, it maps the field’s evolution and underlying research paradigms. The findings reveal a clear shift from early innovation in technical tools toward the realization of clinical value, giving rise to an integrated research system that connects technology, data, clinical practice, and public health. Recent research has progressed beyond initial explorations of medical imaging, standalone devices, and isolated techniques, moving instead toward core domains such as immersive medical visualization, medical data visualization and analytics, health information systems and decision support, AI-assisted epidemic prediction and diagnosis, and integrated IoT-based healthcare frameworks. Looking ahead, an assessment of future trends suggests that, among other directions, the deep integration of explainable artificial intelligence (XAI) with visualization analysis, the development of IoT-driven real-time interactive systems, and the extension of visualization-enabled services from clinical applications toward inclusive population-level health coverage represent core driving forces for the future development of this field. These insights offer strategic guidance for future research, inform the design principles of next-generation visualization systems, and provide new models of interdisciplinary collaboration. The results also offer evidence-based support for health resource planning, technological innovation, and policy formulation. Full article

(This article belongs to the Special Issue Medical Data Visualization)

► Show Figures

Figure 1

Journal Menu

Journal Browser

Information, Volume 17, Issue 3 (March 2026) – 93 articles

Further Information

Guidelines

MDPI Initiatives

Follow MDPI