Information

47 pages, 5755 KB

Open AccessArticle

ZeroDay-LLM: A Large Language Model Framework for Zero-Day Threat Detection in Cybersecurity

by Mohammed Abdullah Alsuwaiket

Information 2025, 16(11), 939; https://doi.org/10.3390/info16110939 (registering DOI) - 28 Oct 2025

Zero-day attacks pose unprecedented challenges to modern cybersecurity frameworks, exploiting unknown vulnerabilities that evade traditional signature-based detection systems. This paper presents ZeroDay-LLM, a novel large language model framework specifically designed for real-time zero-day threat detection in IoT and cloud networks. The proposed system [...] Read more.

Zero-day attacks pose unprecedented challenges to modern cybersecurity frameworks, exploiting unknown vulnerabilities that evade traditional signature-based detection systems. This paper presents ZeroDay-LLM, a novel large language model framework specifically designed for real-time zero-day threat detection in IoT and cloud networks. The proposed system integrates lightweight edge encoders with centralized transformer-based reasoning engines, enabling contextual understanding of network traffic patterns and behavioral anomalies. Through comprehensive evaluation on benchmark cybersecurity datasets including CICIDS2017, NSL-KDD, and UNSW-NB15, ZeroDay-LLM demonstrates superior performance, with a 97.8% accuracy in detecting novel attack signatures, a 23% reduction in false positives compared to traditional intrusion detection systems, and enhanced resilience against adversarial evasion techniques. The framework achieves real-time processing capabilities with an average latency of 12.3 ms per packet analysis while maintaining scalability across heterogeneous network infrastructures. Experimental results across urban, rural, and mixed deployment scenarios validate the practical applicability and robustness of the proposed approach. Full article

(This article belongs to the Special Issue Cyber Security in IoT)

► Show Figures

Graphical abstract

22 pages, 1329 KB

Open AccessArticle

Voices of Researchers: Ethics and Artificial Intelligence in Qualitative Inquiry

by Juan Luis Cabanillas-García, María Cruz Sánchez-Gómez and Irene del Brío-Alonso

Information 2025, 16(11), 938; https://doi.org/10.3390/info16110938 (registering DOI) - 28 Oct 2025

Abstract

The rapid emergence of Generative Artificial Intelligence (GenAI) has sparked a growing debate about its ethical, methodological, and epistemological implications for qualitative research. This study aimed to examine and deeply understand researchers’ perceptions regarding the use of GenAI tools in different phases of [...] Read more.

The rapid emergence of Generative Artificial Intelligence (GenAI) has sparked a growing debate about its ethical, methodological, and epistemological implications for qualitative research. This study aimed to examine and deeply understand researchers’ perceptions regarding the use of GenAI tools in different phases of the qualitative research process. The study involved a sample of 214 researchers from diverse disciplinary areas, with publications indexed in Web of Science or Scopus that apply qualitative methods. Data collection was conducted using an open-ended questionnaire, and analysis was carried out using coding and thematic analysis procedures, which allowed us to identify patterns of perception, user experiences, and barriers. The findings show that, while GenAI is valued for its ability to optimize tasks such as corpus organization, initial coding, transcription, translation, and information synthesis, its implementation raises concerns regarding privacy, consent, authorship, the reliability of results, and the loss of interpretive depth. Furthermore, a dual ecosystem is observed, where some researchers already incorporate it, mainly generative text assistants like ChatGPT, while others have yet to use it or are unfamiliar with it. Overall, the results suggest that the most solid path is an assisted model, supported by clear ethical frameworks, adapted methodological guidelines, and critical training for responsible and humanistic use. Full article

(This article belongs to the Special Issue Generative AI Technologies: Shaping the Future of Higher Education)

► Show Figures

Figure 1

18 pages, 1707 KB

Open AccessArticle

DefAn: Definitive Answer Dataset for LLM Hallucination Evaluation

by A. B. M. Ashikur Rahman, Saeed Anwar, Muhammad Usman, Irfan Ahmad and Ajmal Mian

Information 2025, 16(11), 937; https://doi.org/10.3390/info16110937 (registering DOI) - 28 Oct 2025

Abstract

Large Language Models (LLMs) represent a major step in AI development and are increasingly used in daily applications. However, they are prone to hallucinations, generating claims that contradict established facts, deviating from prompts, and producing inconsistent responses when the same prompt is presented [...] Read more.

Large Language Models (LLMs) represent a major step in AI development and are increasingly used in daily applications. However, they are prone to hallucinations, generating claims that contradict established facts, deviating from prompts, and producing inconsistent responses when the same prompt is presented multiple times. Addressing these issues is challenging due to the lack of comprehensive and easily assessable benchmark datasets. Most existing datasets are limited in scale and scope and rely on multiple-choice questions, which are insufficient for evaluating the generative capabilities of LLMs. To assess hallucination in LLMs, this paper introduces a comprehensive benchmark dataset consisting of over 20,000 unique prompts (more than 75,000 prompts in total) across eight domains. These prompts are designed to elicit definitive, concise, and informative answers. The dataset is divided into two segments: one publicly available for testing and assessing LLM performance, and a hidden segment for benchmarking various LLMs. In our experiments, we tested nine State-of-The-Art (SoTA) models, GPT-4o, GPT-3.5, LLama 2 7B, LLama 3 8B, Gemini 1.0 Pro, Mixtral 8x7B, Zephyr 7B, Deepseek-r1-7b, and Qwen2.5-14B, revealing that overall factual hallucination ranges from 48% to 82% on the public dataset and 31% to 76% on the hidden benchmark. Prompt Misalignment Hallucination ranges up to 95% in the public dataset and up to 94% in the hidden counterpart. Average consistency ranges from 21% to 61% and 44% to 63%, respectively. Domain-wise analysis reveals that LLM performance significantly deteriorates when asked for specific numeric information, whereas it performs moderately with queries involving persons, locations, and dates. Our dataset demonstrates its efficacy and serves as a comprehensive benchmark for evaluating LLM performance. Full article

► Show Figures

Graphical abstract

20 pages, 2200 KB

Open AccessArticle

When Generative AI Goes to the Museum: Visual Stereotyping of Curators and Museum Spaces

by Dirk H. R. Spennemann and Wayne Robinson

Information 2025, 16(11), 936; https://doi.org/10.3390/info16110936 (registering DOI) - 28 Oct 2025

Abstract

Based on 350 visualizations, this paper examines the depiction of museum curators by the popular generative artificial intelligence (AI) model, ChatGPT4o. While the AI-generated representations do not reiterate popular stereotypes of curators as nerdy, conservative in dress, and stuck in time, rummaging through [...] Read more.

Based on 350 visualizations, this paper examines the depiction of museum curators by the popular generative artificial intelligence (AI) model, ChatGPT4o. While the AI-generated representations do not reiterate popular stereotypes of curators as nerdy, conservative in dress, and stuck in time, rummaging through collections, they contrast sharply with real-world demographics. AI-generated imagery severely under-represents women (3.5% vs. 49–72% in reality) and disregards ethnic communities outside of Caucasian communities (0% vs. 18–36%). It not only over-represents young curators (79% vs. approx. 27%) but also renders curators to resemble yuppie professionals or people featured in fashion advertising. Stereotypical attributes are prevalent, with curators widely depicted as having beards and holding clipboards or digital tablets. The findings highlight biases in the generative AI image creation data sets, which are poised to shape an inaccurate portrayal of museum professionals if the images were to be taken uncritically at ‘face value’. Full article

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

► Show Figures

Figure 1

22 pages, 1065 KB

Open AccessArticle

Mapping the Gaze: Comparing the Effectiveness of Bowel-Cancer Screening Advertisements

by Ioanna Yfantidou, Marek Palace, Stefanos Balaskas, Christian Von Wagner, Lee Smith, Brandon May, Jazzine Samuel, Meghna Srivastava, Carlos Santos Barea and Sandro Stoffel

Information 2025, 16(11), 935; https://doi.org/10.3390/info16110935 (registering DOI) - 28 Oct 2025

Abstract

Public-health campaigns have to capture and hold visual attention, but little is known about the influence of message framing and visual appeal on attention to bowel-cancer screening ad campaigns. In a within-subjects test, 42 UK adults aged 40 to 65 viewed 54 static [...] Read more.

Public-health campaigns have to capture and hold visual attention, but little is known about the influence of message framing and visual appeal on attention to bowel-cancer screening ad campaigns. In a within-subjects test, 42 UK adults aged 40 to 65 viewed 54 static adverts that varied by (i) slogan frame—anticipated regret (AR) vs. positive (P); (ii) image type—hand-drawn, older stock, AI-generated; and (iii) identity congruence—viewer ethnicity matched vs. unmatched to the depicted models. Remote eye-tracking measured time to first fixation (TTFF), dwell, fixations, and revisits on a priori pre-defined regions of interest (ROIs); analyses employed linear mixed-effects models (LMMs), generalized estimating equations (GEEs), and median quantile regressions with cluster at the participant level. Across models, the AR slogans produced faster orienting (smaller TTFF) and more intense maintained attention (longer dwell, more fixations and revisits) than the P slogans. Image type set baseline attention (hand-drawn > old stock > AI) but did not significantly decrease the AR benefit, which was equivalent for all visual styles. Identity congruence enhanced early capture (lower TTFF), with small effects for dwell-based measures, suggesting that tailoring benefits only the “first glance.” Anticipated-regret framing is a reliable, design-level alternative to improving both initial capture and sustained processing of screening messages. In practice, the results indicate that advertisers should pair regret-based slogans with warm, human-centred imagery; place slogans in high-salience, low-competition spaces, and, when incorporating AI-generated imagery, reduce composition complexity and exclude uncanny details. These findings ground regret framing as a visual-attention mechanism for public-health campaigns in empirical fact and provide practical recommendations for testing and production. Full article

(This article belongs to the Special Issue Human–Computer Interaction in Marketing: Emerging Interfaces, Cognitive Engagement, and Child–Computer Interaction)

► Show Figures

Figure 1

20 pages, 7297 KB

Open AccessArticle

Student Classroom Behavior Recognition Based on YOLOv8 and Attention Mechanism

by Jingpu Zhang, Lizheng Guo and Xuyang Wang

Information 2025, 16(11), 934; https://doi.org/10.3390/info16110934 (registering DOI) - 27 Oct 2025

Abstract

Accurately recognizing student classroom behaviors is essential for analyzing teacher–student interactions and enabling intelligent educational assessment. Although deep learning offers promising solutions, existing methods often perform poorly in complex classroom environments due to occlusions and subtle, overlapping actions. To address these issues, this [...] Read more.

Accurately recognizing student classroom behaviors is essential for analyzing teacher–student interactions and enabling intelligent educational assessment. Although deep learning offers promising solutions, existing methods often perform poorly in complex classroom environments due to occlusions and subtle, overlapping actions. To address these issues, this article proposes a robust and efficient method for behavior recognition by enhancing the You Only Look Once version 8 (YOLOv8) architecture with a Multi-Head Self-Attention (MHSA) module, termed YOLOv8-MHSA. The integration of MHSA allows the model to capture contextual relationships between distant spatial features, which is critical for distinguishing similar behaviors. For a comprehensive evaluation, we also implement a model with Coordinate Attention (CA). Experimental results on a standard dataset demonstrate the superiority of our YOLOv8-MHSA model, which achieves a precision of 0.86, recall of 0.807, mAP50 of 0.855, and mAP50-95 of 0.677, delivering competitive performance compared to the state-of-the-art SBD-Net. These findings validate that explicit contextual modeling via self-attention significantly boosts performance in fine-grained behavior recognition. Consequently, this research has direct potential applications in providing automated, data-driven tools for teacher training, classroom quality assessment, and, ultimately, supporting the development of personalized education systems. Full article

► Show Figures

Graphical abstract

16 pages, 1629 KB

Open AccessArticle

Planning Future EV Charging Infrastructure by Forecasting Spatio-Temporal Adoption Trends Across Heterogeneous User Segments

by Gheorghe-Daniel Voinea, Florin Gîrbacia, Mihai Duguleană and Cristian-Cezar Postelnicu

Information 2025, 16(11), 933; https://doi.org/10.3390/info16110933 (registering DOI) - 26 Oct 2025

Abstract

The rapid transition to electric vehicles (EVs) requires a charging infrastructure that is both efficient and equitable. Conventional planning approaches, which often deploy chargers in proportion to current EV density, fail to account for the diverse characteristics of EV owners and the evolving [...] Read more.

The rapid transition to electric vehicles (EVs) requires a charging infrastructure that is both efficient and equitable. Conventional planning approaches, which often deploy chargers in proportion to current EV density, fail to account for the diverse characteristics of EV owners and the evolving patterns of adoption across different regions and time periods. This paper introduces an integrated, data-driven framework that addresses these limitations through three stages: segmentation of the EV market, spatio-temporal adoption forecasting for each segment, and optimizing charger placement through a constrained optimization model. The proposed optimization model incorporates equity constraints to ensure minimum service coverage for all user segments while maximizing overall utilization within a fixed budget. Methodologically, the paper contributes a transparent, reproducible framework that unifies user segmentation, geographically resolved adoption forecasting, and an equity-constrained MILP for charger placement. Applying this approach to a dataset of EV registrations in Washington State from 2010 to 2025 and extending it to projections through 2030 demonstrate important improvements in demand coverage. Overall coverage increases from 76.0% to 96.1% compared to a proportional-allocation baseline. More importantly, the proposed framework ensures a minimum of 70% coverage for all user segments. The presented approach is portable to other regions and budget scenarios. These findings show the potential for strategic, data-informed infrastructure planning that balances efficiency and equity, providing actionable insights for policymakers and network operators in the EV transition. Full article

(This article belongs to the Special Issue Artificial Intelligence and Data-Driven Strategies for Advancing Smart, Sustainable, and Resilient Infrastructures and Mobility Systems)

► Show Figures

Figure 1

40 pages, 4004 KB

Open AccessReview

Data Integration and Storage Strategies in Heterogeneous Analytical Systems: Architectures, Methods, and Interoperability Challenges

by Paraskevas Koukaras

Information 2025, 16(11), 932; https://doi.org/10.3390/info16110932 (registering DOI) - 26 Oct 2025

Abstract

In the current scenario of universal accessibility of data, organisations face highly complex challenges related to integrating and processing diverse sets of data in order to meet their analytical needs. This review paper analyses traditional and innovative methods used for data storage and [...] Read more.

In the current scenario of universal accessibility of data, organisations face highly complex challenges related to integrating and processing diverse sets of data in order to meet their analytical needs. This review paper analyses traditional and innovative methods used for data storage and integration, with particular focus on their implications for scalability, consistency, and interoperability within an analytical ecosystem. In particular, it contributes a cross-layer taxonomy linking integration mechanisms (schema matching, entity resolution, and semantic enrichment) to storage/query substrates (row/column stores, NoSQL, lakehouse, and federation), together with comparative tables and figures that synthesise trade-offs and performance/governance levers. Through schema mapping solutions addressing the challenges brought about by structural heterogeneity, storage architectures varying from traditional storage solutions all the way to cloud storage solutions, and ETL pipeline integration using federated query processors, the research provides specific attention for the application of metadata management, with a focus on semantic enrichment using ontologies and lineage management to enable end-to-end traceability and governance. It also covers performance hotspots and caching techniques, along with consistency trade-offs arising out of distributed systems. Empirical case studies from real applications in enterprise lakehouses, scientific exploration activities, and public governance applications serve to invoke this review. Following this work is the possibility of future directions in convergent analytical platforms with support for multiple workloads, along with metadata-centric orchestration with provisions for AI-based integration. Combining technological advancement with practical considerations results in an enabling resource for researchers and practitioners seeking the creation of fault-tolerant, reliable, and future-ready data infrastructure. This review is primarily aimed at researchers, system architects, and advanced practitioners who design and evaluate heterogeneous analytical platforms. It also offers value to graduate students by serving as a structured overview of contemporary methods, thereby bridging academic knowledge with industrial practice. Full article

(This article belongs to the Special Issue Advanced Methods for Multi-Source Information Management, Modeling, and Analysis)

► Show Figures

Graphical abstract

21 pages, 1793 KB

Open AccessArticle

Mining Impersonification Bias in LLMs via Survey Filling

by Marco Bombieri and Marco Rospocher

Information 2025, 16(11), 931; https://doi.org/10.3390/info16110931 (registering DOI) - 26 Oct 2025

Abstract

In this paper, we introduce a survey-based methodology to audit LLM-generated personas by simulating 200 US residents and collecting responses to socio-demographic questions in a zero-shot setting. We investigate whether LLMs default to standardized profiles, how these profiles differ across models, and how [...] Read more.

In this paper, we introduce a survey-based methodology to audit LLM-generated personas by simulating 200 US residents and collecting responses to socio-demographic questions in a zero-shot setting. We investigate whether LLMs default to standardized profiles, how these profiles differ across models, and how conditioning on specific attributes affects the resulting portrayals. Our findings reveal that LLMs often produce homogenized personas that underrepresent demographic diversity and that conditioning on attributes such as gender, ethnicity, or disability may trigger stereotypical shifts. These results highlight implicit biases in LLMs and underscore the need for systematic approaches to evaluate and mitigate fairness risks in model outputs. Full article

(This article belongs to the Special Issue The Role of Artificial Intelligence for Diversity, Equity, and Inclusion)

► Show Figures

Figure 1

25 pages, 6900 KB

Open AccessArticle

Data Mining Archaeogenetic and Linguistic Data Gives an Improved Chronology of the Uralic Language Family

by Peter Z. Revesz

Information 2025, 16(11), 930; https://doi.org/10.3390/info16110930 - 23 Oct 2025

Abstract

Since the early 19th century, linguists have collected enough linguistic data to draw a remarkably stable Uralic language family tree. However, the traditional Uralic language family tree has two main problems. First, it lacks a reliable chronology because linguistic data can suggest that [...] Read more.

Since the early 19th century, linguists have collected enough linguistic data to draw a remarkably stable Uralic language family tree. However, the traditional Uralic language family tree has two main problems. First, it lacks a reliable chronology because linguistic data can suggest that some languages are closer or farther from each other, but that gives only a relative instead of a precise chronology of the branching events. Second, the extinct Mezhovskaya culture in the Ural region and the Minoan civilization on the island of Crete were not incorporated into the Uralic language family, although recent archaeogenetic and linguistic data indicate that their languages also belonged to the Uralic language family. Some recent studies took an essentially purely archaeogenetic approach to the study of the evolution of the Uralic language family. These purely archaeogenetic studies propose linguistically perplexing solutions. This is the first study of the development of the Uralic language family that fully integrates the archaeogenetic, archaeological and linguistic data and proposes a new chronology of the Uralic language family that avoids the above inconsistencies. The new chronology relies on the best current estimates of the formation of the mitochondrial DNA haplogroups that are found among present Uralic language speakers and in samples from various archaeological sites that are associated with Uralic speakers. The new chronology places the various branching events of the Uralic language family tree much earlier than usual, including the split between Proto-Finno-Permic and Proto-Ugric, which is shown to have taken place in the Mesolithic period. The new proposal makes the Bronze Age Minoan language better fit chronologically as well as linguistically into the Uralic language family. Full article

(This article belongs to the Section Information Processes)

► Show Figures

Figure 1

20 pages, 332 KB

Open AccessArticle

Leveraging Machine Learning Techniques to Investigate Media and Information Literacy Competence in Tackling Disinformation

by José Manuel Alcalde-Llergo, Mariana Buenestado Fernández, Carlos Enrique George-Reyes, Andrea Zingoni and Enrique Yeguas-Bolívar

Information 2025, 16(11), 929; https://doi.org/10.3390/info16110929 - 23 Oct 2025

Abstract

This study develops machine learning models to assess Media and Information Literacy (MIL) skills specifically in the context of disinformation among students, particularly future educators and communicators. While the digital revolution has expanded access to information, it has also amplified the spread of [...] Read more.

This study develops machine learning models to assess Media and Information Literacy (MIL) skills specifically in the context of disinformation among students, particularly future educators and communicators. While the digital revolution has expanded access to information, it has also amplified the spread of false and misleading content, making MIL essential for fostering critical thinking and responsible media engagement. Despite its relevance, predictive modeling of MIL in relation to disinformation remains underexplored. To address this gap, a quantitative study was conducted with 723 students in education and communication programs using a validated survey. Classification and regression algorithms were applied to predict MIL competencies and identify key influencing factors. Results show that complex models outperform simpler approaches, with variables such as academic year and prior training significantly improving prediction accuracy. These findings can inform the design of targeted educational interventions and personalized strategies to enhance students’ ability to critically navigate and respond to disinformation in digital environments. Full article

► Show Figures

Figure 1

17 pages, 2618 KB

Open AccessArticle

Optimizer-Aware Fine-Tuning of Whisper Small with Low-Rank Adaption: An Empirical Study of Adam and AdamW

by Hadia Arshad, Tahir Abdullah, Mariam Rehman, Afzaal Hussain, Faria Kanwal and Mehwish Parveen

Information 2025, 16(11), 928; https://doi.org/10.3390/info16110928 - 22 Oct 2025

Abstract

Whisper is a transformer-based multilingual model that has illustrated state-of-the-art behavior in numerous languages. However, the efficiency remains persistent with the limited computational resources. To address this issue, an experiment was performed on librispeech-train-clean-100 for training purposes. The test-clean set was utilized to [...] Read more.

Whisper is a transformer-based multilingual model that has illustrated state-of-the-art behavior in numerous languages. However, the efficiency remains persistent with the limited computational resources. To address this issue, an experiment was performed on librispeech-train-clean-100 for training purposes. The test-clean set was utilized to evaluate its performance. To enhance efficiency and to cater the computational needs, a parameter-efficient fine-tuning technique, i.e., Low-Rank Adaptation, was employed to add a limited number of trainable parameters into the frozen layers of the model. The results showed that Low-Rank Adaptation attained excellent Automatic Speech Recognition results while using fewer computational resources, showing its effectiveness for resource-saving adaptation. The research work emphasizes the promise of Low-Rank Adaptation as a lightweight and scalable fine-tuning strategy for large speech models using a transformer architecture. The baseline Whisper Small model achieved a word error rate of 16.7% without any parameter-efficient adaptation. In contrast, the Low-Rank Adaptation enhanced fine-tuned model achieved a lower word error rate of 6.08%, demonstrating the adaptability of the proposed parameter-efficient approach. Full article

► Show Figures

Figure 1

32 pages, 3458 KB

Open AccessArticle

Architecture Building Blocks for Data Governance in Data Spaces

by Marta Zorrilla and Juan Yebenes

Information 2025, 16(11), 927; https://doi.org/10.3390/info16110927 - 22 Oct 2025

Abstract

The growing importance of data as a driver of the digital economy is promoting the creation of data spaces for the secure and controlled exchange of data between organizations. Data governance is emerging as an essential pillar to ensure efficient, ethical and transparent [...] Read more.

The growing importance of data as a driver of the digital economy is promoting the creation of data spaces for the secure and controlled exchange of data between organizations. Data governance is emerging as an essential pillar to ensure efficient, ethical and transparent access and use of data in these ecosystems. The article reviews the state of the art to identify the specific requirements that data governance must address in data spaces and proposes a reference enterprise architecture to facilitate the design, development and implementation of a data governance system for a data space scenario. The proposed framework has already been formally defined and validated in the context of Industry 4.0, and is now adapted to the particular characteristics and needs of data spaces. This architecture focuses on key aspects of data governance in data spaces, such as new requirements, principles, organization, roles and responsibilities, and data quality, security and metadata management, as well as the data lifecycle in the data space. This research contributes to guiding data space government bodies to formalize data strategies and high-level governance principles in concrete architectural components that establish the capacities to be implemented within the data ecosystem. To support practical adoption, this work also provides clarifying examples of different blocks of architecture. Full article

► Show Figures

Graphical abstract

45 pages, 2089 KB

Open AccessArticle

PEARL: A Rubric-Driven Multi-Metric Framework for LLM Evaluation

by Catalin Anghel, Andreea Alexandra Anghel, Emilia Pecheanu, Marian Viorel Craciun, Adina Cocu and Cristian Niculita

Information 2025, 16(11), 926; https://doi.org/10.3390/info16110926 - 22 Oct 2025

Abstract

Background and objectives: Evaluating Large Language Models (LLMs) presents two interrelated challenges: the general problem of assessing model performance across diverse tasks and the specific problem of using LLMs themselves as evaluators in pedagogical and educational contexts. Existing approaches often rely on single [...] Read more.

Background and objectives: Evaluating Large Language Models (LLMs) presents two interrelated challenges: the general problem of assessing model performance across diverse tasks and the specific problem of using LLMs themselves as evaluators in pedagogical and educational contexts. Existing approaches often rely on single metrics or opaque preference-based methods, which fail to capture critical dimensions such as explanation quality, robustness, and argumentative diversity—attributes essential in instructional settings. This paper introduces PEARL, a novel framework conceived, operationalized, and evaluated in the present work using LLM-based scorers, designed to provide interpretable, reproducible, and pedagogically meaningful assessments across multiple performance dimensions. Methods: PEARL integrates three specialized rubrics—Technical, Argumentative, And Explanation-focused—covering aspects such as factual accuracy, clarity, completeness, originality, dialecticality, and explanatory usefulness. The framework defines seven complementary metrics: Rubric Win Count (RWC), Global Win Rate (GWR), Rubric Mean Advantage (RMA), Consistency Spread (CS), Win Confidence Score (WCS), Explanation Quality Index (EQI), and Dialectical Presence Rate (DPR). We evaluated PEARL by evaluating eight open-weight instruction-tuned LLMs across 51 prompts, with outputs scored independently by GPT-4 and LLaMA 3:instruct. This constitutes LLM-based evaluation, and observed alignment with the GPT-4 proxy is mixed across metrics. Results: Preference-based metrics (RMA, RWC, and GWR) show evidence of group separation, reported with bootstrap confidence intervals and interpreted as exploratory due to small samples, while robustness-oriented (CS and WCS) and reasoning-diversity (DPR) metrics capture complementary aspects of performance not reflected in global win rate. RMA and RWC exhibit statistically significant, FDR-controlled correlations with the GPT-4 proxy, and correlation mapping highlights the complementary and partially orthogonal nature of PEARL’s evaluation dimensions. Originality: PEARL is the first LLM evaluation framework to combine multi-rubric scoring, explanation-aware metrics, robustness analysis, and multi-LLM-evaluator analysis into a single, extensible system. Its multidimensional design supports both high-level benchmarking and targeted diagnostic assessment, offering a rigorous, transparent, and versatile methodology for researchers, developers, and educators working with LLMs in high-stakes and instructional contexts. Full article

► Show Figures

Figure 1

33 pages, 4858 KB

Open AccessArticle

Multi-Criteria Assessment: A Case Study Integrating Eco-design Principles in Sustainable Manufacturing

by Khadija Sarquah, Caitlin Walls, Marta Revello, Maja Jelić and Gesa Beck

Information 2025, 16(11), 925; https://doi.org/10.3390/info16110925 - 22 Oct 2025

Abstract

This study integrates Eco-design principles and the Life Cycle approach in an MCA to evaluate the sustainability performance of manufacturing routes. The assessment is applied to conventional production across five use cases involving complex geometry parts. The aim is to evaluate areas of [...] Read more.

This study integrates Eco-design principles and the Life Cycle approach in an MCA to evaluate the sustainability performance of manufacturing routes. The assessment is applied to conventional production across five use cases involving complex geometry parts. The aim is to evaluate areas of material criticality, environmental impacts, chemical risks, as well as social aspects, including gender dimensions (C-MET-ESG). Outcomes are synthesised into colour-coded hotspot tables and Eco-design recommendations. Key findings highlight opportunities such as substituting high-criticality alloys, increasing material efficiency, and promoting gender inclusive workplace practices. Technological transitions from CNC machining and hazardous post-processing to laser and additive manufacturing further enhance safety, resource efficiency, and resilience. The novelty of this study lies in the integration of LCA principles, the C-MET-ESG matrix, and CRA-SSbD guidelines within an MCA, establishing a hazard-aware, socially inclusive, and technically robust framework. This approach provides life cycle linked evidence that connects early design choices to sustainability outcomes. Furthermore, the study offers a transferable methodology for sustainable manufacturing in both established and emerging technologies. Full article

(This article belongs to the Special Issue New Applications in Multiple Criteria Decision Analysis, 3rd Edition)

► Show Figures

Figure 1

24 pages, 10501 KB

Open AccessArticle

Unveiling Dark Web Identity Patterns: A Network-Based Analysis of Identification Types and Communication Channels in Illicit Activities

by Luis de-Marcos, Adrián Domínguez-Díaz, Javier Junquera-Sánchez, Carlos Cilleruelo and José-Javier Martínez-Herráiz

Information 2025, 16(11), 924; https://doi.org/10.3390/info16110924 - 22 Oct 2025

Abstract

The Dark Web, a hidden segment of the internet, has become a hub for illicit activities, facilitated by various forms of digital identification (IDs) such as email addresses, Telegram accounts, and cryptocurrency wallets. This study conducts a comprehensive analysis of the Dark Web’s [...] Read more.

The Dark Web, a hidden segment of the internet, has become a hub for illicit activities, facilitated by various forms of digital identification (IDs) such as email addresses, Telegram accounts, and cryptocurrency wallets. This study conducts a comprehensive analysis of the Dark Web’s identification and communication patterns, focusing on the roles of different ID types and their associated activities. Using a dataset of Dark Web documents, we construct and analyze a bipartite network to model the relationships between IDs and web documents, employing graph–theoretical metrics such as degree centrality, closeness centrality, betweenness centrality, and k-core decomposition, while analyzing subnetworks formed by ID type. Our findings reveal that Telegram forms the backbone of the network, serving as the primary communication tool for hacking-related activities, particularly within Russian-speaking communities. In contrast, email plays a more decentralized role, facilitating finance–crypto and other activities but with a high level of fragmentation and English as the predominant language. XMR (Monero) wallets emerge as a key component in financial transactions, forming a cohesive subnetwork focused on cryptocurrency-related activities. The analysis also highlights the modular and hierarchical nature of the Dark Web, with distinct clusters for hacking, finance–crypto, and drugs–narcotics, often operating independently but with some cross-topic interactions. This study provides a foundation for understanding the Dark Web’s structure and dynamics, offering insights that can inform strategies for monitoring and mitigating its risks. Full article

(This article belongs to the Section Information Security and Privacy)

► Show Figures

Graphical abstract

3 pages, 153 KB

Open AccessCorrection

Correction: Wu et al. Critical Factors for Predicting Users’ Acceptance of Digital Museums for Experience-Influenced Environments. Information 2021, 12, 426

by Yue Wu, Qianling Jiang, Shiyu Ni and Hui’e Liang

Information 2025, 16(11), 923; https://doi.org/10.3390/info16110923 - 22 Oct 2025

Abstract

There was an error in the original publication [...] Full article

21 pages, 298 KB

Open AccessArticle

Video Games in Schools: Putting Flow State in Context

by Marcello Sarini, Francesco Bocci, Giulia Centini, Anna Maso and Luca Pianigiani

Information 2025, 16(10), 922; https://doi.org/10.3390/info16100922 - 21 Oct 2025

Abstract

There is growing interest in exploring the positive effects of commercial video games for educational purposes. These investigations focus on enhancing learning outcomes but also on improving emotional and physical well-being, which may lead to greater motivation to learn and more positive learning [...] Read more.

There is growing interest in exploring the positive effects of commercial video games for educational purposes. These investigations focus on enhancing learning outcomes but also on improving emotional and physical well-being, which may lead to greater motivation to learn and more positive learning experiences. However, few studies focus on the role of flow state in these phenomena. In line with this, we present two studies, the former conducted in a primary school in Northern Italy, and the latter in two secondary schools in Central Italy. Both of them aim to investigate the hidden effects of gaming at school in connection with the flow state. The first study, involving students from four classes, aimed at revealing collective dynamics unique to each class. The second study involved students from three classes of two different schools. This study investigated the peculiarity of the effects of different genres of games played in classes. We observed how flow state in context raised meaningful differences in some of the variables measured. These variations seem to reflect distinctive traits regarding playing in general and in the specifics of each class. These findings suggest that educators could consider the unique characteristics that gaming makes evident when designing educational strategies, potentially tailoring the learning process to better align with the specific dynamics of the context. Full article

► Show Figures

Figure 1

22 pages, 5641 KB

Open AccessArticle

A Globally Optimal Alternative to MLP

by Zheng Li, Jerry Cheng and Huanying Helen Gu

Information 2025, 16(10), 921; https://doi.org/10.3390/info16100921 - 21 Oct 2025

Abstract

In deep learning, achieving the global minimum poses a significant challenge, even for relatively simple architectures such as Multi-Layer Perceptrons (MLPs). To address this challenge, we visualized model states at both local and global optima, thereby identifying the factors that impede the transition [...] Read more.

In deep learning, achieving the global minimum poses a significant challenge, even for relatively simple architectures such as Multi-Layer Perceptrons (MLPs). To address this challenge, we visualized model states at both local and global optima, thereby identifying the factors that impede the transition of models from local to global minima when employing conventional model training methodologies. Based on these insights, we propose the Lagrange Regressor (LReg), a framework that is mathematically equivalent to MLPs. Rather than updates via optimization techniques, LReg employs a Mesh-Refinement–Coarsening (discrete) process to ensure the convergence of the model’s loss function to the global minimum. LReg achieves faster convergence and overcomes the inherent limitations of neural networks in fitting multi-frequency functions. Experiments conducted on large-scale benchmarks including ImageNet-1K (image classification), GLUE (natural language understanding), and WikiText (language modeling) show that LReg consistently enhances the performance of pre-trained models, significantly lowers test loss, and scales effectively to big data scenarios. These results underscore LReg’s potential as a scalable, optimization-free alternative for deep learning in large and complex datasets, aligning closely with the goals of innovative big data analytics. Full article

(This article belongs to the Special Issue Machine Learning and Data Mining: Innovations in Big Data Analytics, 2nd Edition)

► Show Figures

Figure 1

Journal Description

Information

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Topical Collections

Further Information

Guidelines

MDPI Initiatives

Follow MDPI