Information

20 pages, 613 KiB

Open AccessArticle

Max-Min Secrecy Rate for UAV-Assisted Energy Harvesting IoT Networks

by Mingrui Zheng, Tianrui Feng and Tengjiao He

Information 2025, 16(2), 158; https://doi.org/10.3390/info16020158 - 19 Feb 2025

Viewed by 471

The future Internet of Things (IoT) will consist of energy harvesting devices and Unmanned Aerial Vehicles (UAVs) to support applications in remote areas. However, as UAVs communicate with IoT devices using broadcast channels, information leakage emerges as a critical security threat. This paper [...] Read more.

The future Internet of Things (IoT) will consist of energy harvesting devices and Unmanned Aerial Vehicles (UAVs) to support applications in remote areas. However, as UAVs communicate with IoT devices using broadcast channels, information leakage emerges as a critical security threat. This paper considers the problem of maximizing the minimum secrecy rate in an energy harvesting IoT network supported by two UAVs, where one acts as a server to collect data from devices, and the other is an eavesdropper to intercept data transmission. It presents a novel Mixed-Integer Nonlinear Program (MINLP), which we then linearize into a Mixed-Integer Linear Program (MILP) problem. It also proposes a heuristic solution called Fly Nearest Location (FNL). Both solutions determine (i) the UAV server’s flight routing, flight time, and computation time, as well as (ii) the energy usage and operation mode of IoT devices. Our results show that FNL achieves on average 78.15% of MILP’s performance. Full article

► Show Figures

Graphical abstract

63 pages, 22670 KiB

Open AccessReview

Style Transfer Review: Traditional Machine Learning to Deep Learning

by Yao Xu, Min Xia, Kai Hu, Siyi Zhou and Liguo Weng

Information 2025, 16(2), 157; https://doi.org/10.3390/info16020157 - 19 Feb 2025

Viewed by 2726

Abstract

Style transfer is a technique that learns style features from different domains and applies these features to other images. It can not only play a role in the field of artistic creation but also has important significance in image processing, video processing, and [...] Read more.

Style transfer is a technique that learns style features from different domains and applies these features to other images. It can not only play a role in the field of artistic creation but also has important significance in image processing, video processing, and other fields. However, at present, style transfer still faces some challenges, such as the balance between style and content, the model generalization ability, and diversity. This article first introduces the origin and development process of style transfer and provides a brief overview of existing methods. Next, this article explores research work related to style transfer, introduces some metrics used to evaluate the effect of style transfer, and summarizes datasets. Subsequently, this article focuses on the application of the currently popular deep learning technology for style transfer and also mentions the application of style transfer in video. Finally, the article discusses possible future directions for this field. Full article

(This article belongs to the Special Issue Surveys in Information Systems and Applications)

► Show Figures

Figure 1

22 pages, 4188 KiB

Open AccessArticle

Video-Based Information Mediating Opportunities for Professional Development: A Research Intervention with Teaching-Focused Lecturers in Higher Education

by Philip Moffitt and Brett Bligh

Information 2025, 16(2), 156; https://doi.org/10.3390/info16020156 - 19 Feb 2025

Viewed by 465

Abstract

This paper contributes to the growing international interest in using video-based information in education, training, and professional development. It describes an empirical study in which we analyse the use of video-based information as educational practitioners negotiate, design, and enact their own professional development. [...] Read more.

This paper contributes to the growing international interest in using video-based information in education, training, and professional development. It describes an empirical study in which we analyse the use of video-based information as educational practitioners negotiate, design, and enact their own professional development. In our study, participants are teaching-focused lecturers in engineering higher education. We describe a research intervention using the Change Laboratory methodology, with expansive learning, where video-based information is embroiled throughout. Our analyses show that video acts at various epistemic levels, from using video-based information to support claims to truth to using video-based information to provoke social negotiations of the partiality of knowledge. We examine how video-based information acts as a mediating technology for imagining, negotiating, and reflexively implementing professional developmental intentions. Our core argument is that practitioners can benefit from understanding how video-based information can mediate their own professional development in relational ways. We make three substantive contributions to scholarship: evincing a need for the prioritisation of understanding diverse epistemic functions of video-based information, advancing understanding of the role of video in theoretically informed social negotiation, and exemplifying methodological arrangements that move video-based information beyond visual representation. Full article

(This article belongs to the Special Issue Feature Papers in Information in 2024–2025)

► Show Figures

Figure 1

26 pages, 3721 KiB

Open AccessArticle

Schema Understandability: A Comprehensive Empirical Study of Requirements Metrics

by Tanu Singh, Vinod Patidar, Manu Singh and Álvaro Rocha

Information 2025, 16(2), 155; https://doi.org/10.3390/info16020155 - 19 Feb 2025

Viewed by 994

Abstract

Ensuring high-quality data warehouses is crucial for organizations, as they provide the reliable information needed for informed decision-making. While various methodologies emphasize the importance of requirements, conceptual, logical, and physical models in developing data warehouses, empirical quality assessment of these models remains underexplored, [...] Read more.

Ensuring high-quality data warehouses is crucial for organizations, as they provide the reliable information needed for informed decision-making. While various methodologies emphasize the importance of requirements, conceptual, logical, and physical models in developing data warehouses, empirical quality assessment of these models remains underexplored, especially requirements models. To bridge this gap, this study focuses on assessment of requirements metrics for predicting the understandability of requirements schemas, a key indicator of model quality. In this empirical study, 28 requirements schemas were classified into understandable and non-understandable clusters using the k-means clustering technique. The study then employed six classification techniques—logistic regression, naive Bayes, linear discriminant analysis with decision tree, reinforcement learning, voting rule, and a hybrid approach—within both univariate and multivariate models to identify strong predictors of schema understandability. Results indicate that 13 out of 17 requirements metrics are robust predictors of schema understandability. Furthermore, a comparative performance analysis of the classification techniques reveals that the hybrid classifier outperforms other techniques across key evaluation parameters, including accuracy, sensitivity, specificity, and AUC. These findings highlight the potential of requirements metrics as effective predictors of schema understandability, contributing to improved quality assessment and the development of better conceptual data models for data warehouses. Full article

(This article belongs to the Special Issue Editorial Board Members’ Collection Series: "Information Systems")

► Show Figures

Figure 1

33 pages, 3144 KiB

Open AccessArticle

CNN-Based Optimization for Fish Species Classification: Tackling Environmental Variability, Class Imbalance, and Real-Time Constraints

by Amirhosein Mohammadisabet, Raza Hasan, Vishal Dattana, Salman Mahmood and Saqib Hussain

Information 2025, 16(2), 154; https://doi.org/10.3390/info16020154 - 19 Feb 2025

Viewed by 633

Abstract

Automated fish species classification is essential for marine biodiversity monitoring, fisheries management, and ecological research. However, challenges such as environmental variability, class imbalance, and computational demands hinder the development of robust classification models. This study investigates the effectiveness of convolutional neural network (CNN)-based [...] Read more.

Automated fish species classification is essential for marine biodiversity monitoring, fisheries management, and ecological research. However, challenges such as environmental variability, class imbalance, and computational demands hinder the development of robust classification models. This study investigates the effectiveness of convolutional neural network (CNN)-based models and hybrid approaches to address these challenges. Eight CNN architectures, including DenseNet121, MobileNetV2, and Xception, were compared alongside traditional classifiers like support vector machines (SVMs) and random forest. DenseNet121 achieved the highest accuracy (90.2%), leveraging its superior feature extraction and generalization capabilities, while MobileNetV2 balanced accuracy (83.57%) with computational efficiency, processing images in 0.07 s, making it ideal for real-time deployment. Advanced preprocessing techniques, such as data augmentation, turbidity simulation, and transfer learning, were employed to enhance dataset robustness and address class imbalance. Hybrid models combining CNNs with traditional classifiers achieved intermediate accuracy with improved interpretability. Optimization techniques, including pruning and quantization, reduced model size by 73.7%, enabling real-time deployment on resource-constrained devices. Grad-CAM visualizations further enhanced interpretability by identifying key image regions influencing predictions. This study highlights the potential of CNN-based models for scalable, interpretable fish species classification, offering actionable insights for sustainable fisheries management and biodiversity conservation. Full article

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

► Show Figures

Graphical abstract

33 pages, 3771 KiB

Open AccessReview

Understanding Social Engineering Victimisation on Social Networking Sites: A Comprehensive Review of Factors Influencing User Susceptibility to Cyber-Attacks

by Saad S. Alshammari, Ben Soh and Alice Li

Information 2025, 16(2), 153; https://doi.org/10.3390/info16020153 - 19 Feb 2025

Viewed by 543

Abstract

The widespread adoption of social networking sites (SNSs) has brought social-engineering victimisation (SEV) to the forefront as a significant concern in recent years. Common examples of social-engineering attacks include phishing websites, fake user accounts, fraudulent messages, impersonation of close friends, and malicious links [...] Read more.

The widespread adoption of social networking sites (SNSs) has brought social-engineering victimisation (SEV) to the forefront as a significant concern in recent years. Common examples of social-engineering attacks include phishing websites, fake user accounts, fraudulent messages, impersonation of close friends, and malicious links shared through comments or posts on SNS platforms. The increasing number of SNS users is closely linked to a rise in SEV incidents. Consequently, it is essential to explore relevant theories, frameworks, and contributing factors to better understand this phenomenon. This study systematises and analyses 47 scholarly works on SEV in SNSs, examining theories, frameworks, and influencing factors. A total of 90 independent variables were identified and grouped into seven perspectives: socio-demographics, personality traits, socio-emotional factors, habitual factors, perceptual/cognitive factors, message characteristics, and sender characteristics; these were considered alongside mediating variables. The correlations between these variables and victimisation outcomes were evaluated, uncovering factors that increase vulnerability and highlighting contradictory findings in existing studies. This systematised analysis emphasises the limitations in current research and identifies future research directions in order to deepen the understanding of the factors influencing SEV. By addressing these gaps, this study aims to advance mitigation strategies and provide actionable insights to reduce SEV in SNS contexts. Full article

(This article belongs to the Special Issue Recent Developments and Implications in Web Analysis)

► Show Figures

Figure 1

15 pages, 3085 KiB

Open AccessArticle

Early Detection of Skin Diseases Across Diverse Skin Tones Using Hybrid Machine Learning and Deep Learning Models

by Akasha Aquil, Faisal Saeed, Souad Baowidan, Abdullah Marish Ali and Nouh Sabri Elmitwally

Information 2025, 16(2), 152; https://doi.org/10.3390/info16020152 - 19 Feb 2025

Viewed by 898

Abstract

Skin diseases in melanin-rich skin often present diagnostic challenges due to the unique characteristics of darker skin tones, which can lead to misdiagnosis or delayed treatment. This disparity impacts millions within diverse communities, highlighting the need for accurate, AI-based diagnostic tools. In this [...] Read more.

Skin diseases in melanin-rich skin often present diagnostic challenges due to the unique characteristics of darker skin tones, which can lead to misdiagnosis or delayed treatment. This disparity impacts millions within diverse communities, highlighting the need for accurate, AI-based diagnostic tools. In this paper, we investigated the performance of three machine learning methods -Support Vector Machines (SVMs), Random Forest (RF), and Decision Trees (DTs)-combined with state-of-the-art (SOTA) deep learning models, EfficientNet, MobileNetV2, and DenseNet121, for predicting skin conditions using dermoscopic images from the HAM10000 dataset. The features were extracted using the deep learning models, with the labels encoded numerically. To address the data imbalance, SMOTE and resampling techniques were applied. Additionally, Principal Component Analysis (PCA) was used for feature reduction, and fine-tuning was performed to optimize the models. The results demonstrated that RF with DenseNet121 achieved a superior accuracy of 98.32%, followed by SVM with MobileNetV2 at 98.08%, and Decision Tree with MobileNetV2 at 85.39%. The proposed methods overcome the SVM with the SOTA EfficientNet model, validating the robustness of the proposed approaches. Evaluation metrics such as accuracy, precision, recall, and F1-score were used to benchmark performance, showcasing the potential of these methods in advancing skin disease diagnostics for diverse populations. Full article

(This article belongs to the Special Issue AI-Based Image Processing and Computer Vision)

► Show Figures

Figure 1

26 pages, 345 KiB

Open AccessReview

Popularity Bias in Recommender Systems: The Search for Fairness in the Long Tail

by Filippo Carnovalini, Antonio Rodà and Geraint A. Wiggins

Information 2025, 16(2), 151; https://doi.org/10.3390/info16020151 - 19 Feb 2025

Viewed by 1459

Abstract

The importance of recommender systems has grown in recent years, as these systems are becoming one of the primary ways in which we access content on the Internet. Along with their use, concerns about the fairness of the recommendations they propose have rightfully [...] Read more.

The importance of recommender systems has grown in recent years, as these systems are becoming one of the primary ways in which we access content on the Internet. Along with their use, concerns about the fairness of the recommendations they propose have rightfully risen. Recommender systems are known to be affected by popularity bias, the disproportionate preference towards popular items. While this bias stems from human tendencies, algorithms used in recommender systems can amplify it, resulting in unfair treatment of end-users and/or content creators. This article proposes a narrative review of the relevant literature to characterize and understand this phenomenon, both in human and algorithmic terms. The analysis of the literature highlighted the main themes and underscored the need for a multi-disciplinary approach that examines the interplay between human cognition, algorithms, and socio-economic factors. In particular, the article discusses how the overall fairness of recommender systems is impacted by popularity bias. We then describe the approaches that have been used to mitigate the harmful effects of this bias and discuss their effectiveness in addressing the issue, finding that some of the current approaches fail to face the problem in its entirety. Finally, we identify some open problems and research opportunities to help the advancement of research in the fairness of recommender systems. Full article

(This article belongs to the Special Issue Applications of Information Extraction, Knowledge Graphs, and Large Language Models)

21 pages, 2248 KiB

Open AccessArticle

AI vs. Human-Authored Headlines: Evaluating the Effectiveness, Trust, and Linguistic Features of ChatGPT-Generated Clickbait and Informative Headlines in Digital News

by Vasile Gherheș, Marcela Alina Fărcașiu, Mariana Cernicova-Buca and Claudiu Coman

Information 2025, 16(2), 150; https://doi.org/10.3390/info16020150 - 18 Feb 2025

Viewed by 1029

Abstract

This study explores possible applications of AI technology in online journalism, given the predictions that speed and adaptation to the new medium will increase the penetration of automation in the production business. The literature shows that while the human supervision of journalistic workflow [...] Read more.

This study explores possible applications of AI technology in online journalism, given the predictions that speed and adaptation to the new medium will increase the penetration of automation in the production business. The literature shows that while the human supervision of journalistic workflow is still considered vital, the journalistic workflow is changing in nature, with the writing of micro-content being entrusted to ChatGPT-3.5 among the most visible features. This research assesses readers’ reactions to different headline styles as tested on a sample of 624 students from Timisoara, Romania, asked to evaluate the qualities of a mix of human-written vs. AI-generated headlines. The results show that AI-generated, informative headlines were perceived by more than half of the respondents as the most trustworthy and representative of the media content. Clickbait headlines, regardless of their source, were considered misleading and rated as manipulative (44.7%). In addition, 54.5% of respondents reported a decrease in trust regarding publications that frequently use clickbait techniques. A linguistic analysis was conducted to grasp the qualities of the headlines that triggered the registered responses. This study provides insights into the potential of AI-enabled tools to reshape headline writing practices in digital journalism. Full article

(This article belongs to the Special Issue Advances in Human-Centered Artificial Intelligence)

► Show Figures

Figure 1

21 pages, 2256 KiB

Open AccessArticle

Optimization of Worker Redeployment for Enhancing Customer Service Performance

by Hyunho Kim, Wonseok Kang and Eunmi Lee

Information 2025, 16(2), 149; https://doi.org/10.3390/info16020149 - 18 Feb 2025

Viewed by 267

Abstract

This study considers ways in which workers can be allocated dynamically in the few hours before the truck departure to flush the system of orders that are almost completed, thereby increasing the service performance of the system. To implement a worker allocation policy [...] Read more.

This study considers ways in which workers can be allocated dynamically in the few hours before the truck departure to flush the system of orders that are almost completed, thereby increasing the service performance of the system. To implement a worker allocation policy correctly, we need to answer the following questions: how many workers should we move, and when? We present the optimal number of workers required and the switching time for the proposed three dynamic worker reallocation policies through simulation experiments. The number of workers required was determined by the difference between the current and target probability of success of an order in the system based on the state-dependent sojourn time distribution, and the performance of the system was measured by Next Scheduled Departure (NSD). We find that the policies with late switching times and higher target probability of success have a greater effect on customer satisfaction. Our results suggest that it is possible to improve service performance significantly, in some conditions, by moving the right number of workers to the right place at the right time. Full article

(This article belongs to the Section Information and Communications Technology)

► Show Figures

Figure 1

20 pages, 5332 KiB

Open AccessArticle

An Adaptive Fatigue Detection Model for Virtual Reality-Based Physical Therapy

by Sergio Martinez-Cid, Mohamed Essalhi, Vanesa Herrera, Javier Albusac, Santiago Schez-Sobrino and David Vallejo

Information 2025, 16(2), 148; https://doi.org/10.3390/info16020148 - 17 Feb 2025

Viewed by 547

Abstract

This paper introduces a fatigue detection model specifically designed for immersive virtual reality (VR) environments, aimed at facilitating upper limb rehabilitation for individuals with spinal cord injuries (SCIs). The model’s primary application centers on the Box-and-Block Test, providing healthcare professionals with a reliable [...] Read more.

This paper introduces a fatigue detection model specifically designed for immersive virtual reality (VR) environments, aimed at facilitating upper limb rehabilitation for individuals with spinal cord injuries (SCIs). The model’s primary application centers on the Box-and-Block Test, providing healthcare professionals with a reliable tool to monitor patient progress and adapt rehabilitation routines. At its core, the model employs data fusion techniques via ordered weighted averaging (OWA) operators to aggregate multiple metrics captured by the VR rehabilitation system. Additionally, fuzzy logic is employed to personalize fatigue assessments. Therapists are provided with a detailed classification of fatigue levels alongside a video-based visual representation that highlights critical moments of fatigue during the exercises. The experimental methodology involved testing the fatigue detection model with both healthy participants and patients, using immersive VR-based rehabilitation scenarios and validating its accuracy through self-reported fatigue levels and therapist observations. Furthermore, the model’s scalable design promotes its integration into remote rehabilitation systems, highlighting its adaptability to diverse clinical scenarios and its potential to enhance accessibility to rehabilitation services. Full article

(This article belongs to the Special Issue Advances in Human-Centered Artificial Intelligence)

► Show Figures

Figure 1

29 pages, 6722 KiB

Open AccessArticle

Framework for Addressing Imbalanced Data in Aviation with Federated Learning

by Igor Kabashkin

Information 2025, 16(2), 147; https://doi.org/10.3390/info16020147 - 16 Feb 2025

Viewed by 676

Abstract

The aviation industry generates vast amounts of data across multiple stakeholders, but critical faults and anomalies occur rarely, creating inherently imbalanced datasets that complicate machine learning applications. Traditional centralized approaches are further constrained by privacy concerns and regulatory requirements that limit data sharing [...] Read more.

The aviation industry generates vast amounts of data across multiple stakeholders, but critical faults and anomalies occur rarely, creating inherently imbalanced datasets that complicate machine learning applications. Traditional centralized approaches are further constrained by privacy concerns and regulatory requirements that limit data sharing among stakeholders. This paper presents a novel framework for addressing imbalanced data challenges in aviation through federated learning, focusing on fault detection, predictive maintenance, and safety management. The proposed framework combines specialized techniques for handling imbalanced data with privacy-preserving federated learning to enable effective collaboration while maintaining data security. The framework incorporates local resampling methods, cost-sensitive learning, and weighted aggregation mechanisms to improve minority class detection performance. The framework is validated through extensive experiments involving multiple aviation stakeholders, demonstrating a 23% improvement in fault detection accuracy and a 17% reduction in remaining useful life prediction error compared to conventional models. Results show the enhanced detection of rare but critical faults, improved maintenance scheduling accuracy, and effective risk assessment across distributed aviation datasets. The proposed framework provides a scalable and practical solution for using distributed aviation data while addressing both class imbalance and privacy concerns, contributing to improved safety and operational efficiency in the aviation industry. Full article

(This article belongs to the Special Issue Machine Learning Approaches for Imbalanced Domains: Emerging Trends and Applications)

► Show Figures

Graphical abstract

31 pages, 4660 KiB

Open AccessArticle

Measuring Innovation Potential in Ecuadorian ICT Companies: Development and Application of the CRI-IRT Model

by Christian Anasi, Andrés Robalino-López, Verónica Morales and Carlos Almeida

Information 2025, 16(2), 146; https://doi.org/10.3390/info16020146 - 16 Feb 2025

Viewed by 453

Abstract

Innovation is recognized as a key source of competitive advantage for organizations and a driver of societal well-being. Therefore, managing and quantifying innovation is necessary to fully leverage its potential. Although many innovation measurement tools exist, contextual differences limit their applicability, highlighting the [...] Read more.

Innovation is recognized as a key source of competitive advantage for organizations and a driver of societal well-being. Therefore, managing and quantifying innovation is necessary to fully leverage its potential. Although many innovation measurement tools exist, contextual differences limit their applicability, highlighting the need for tools tailored to regional and national specificities. This study aimed to develop a measurement tool for the Ecuadorian context, utilizing the Capacities, Results, and Impacts (CRI) model questionnaire for data collection and Item Response Theory (IRT) as a processing method. The CRI model offers a comprehensive framework for assessing innovation potential, enabling a dynamic understanding of this potential while highlighting nuances specific to regional contexts, particularly in Ecuador. Complementing this, IRT—a statistical framework for measuring latent traits—offers several advantages over classical methodologies, such as Classical Test Theory (CTT) and simple aggregate scoring methods. Unlike classical approaches, which often lack precision at extreme ability levels and are heavily sample-dependent, IRT provides item-level analysis, ensures parameter invariance across samples, and maintains accuracy across a wide range of latent trait levels. Together, these methodologies ensure a highly reliable and context-specific innovation measurement tool. Six IRT models were fitted using data from a national multisector sample of 321 organizations, providing a multidimensional measurement scale specialized in measuring innovation potential in Ecuador. These scales were later applied in a case study on the Information and Communication Technology (ICT) sector in Quito, Ecuador. The findings showed that ICT organizations had higher innovation potential than other industries in Ecuador but faced weaknesses in areas like access to funding. Based on these results, targeted strategies were proposed to address these weaknesses and foster innovation within the ICT sector. This research contributes to the field of managing innovation in Ecuador and the broader Latin American region by developing a context-specific and adaptable tool to benchmark innovation, guide organizational strategies, and shape public policy. While the ICT sector was identified as a key driver for addressing Ecuador’s innovation challenges, this research provides valuable contributions toward tackling these challenges and fostering innovation in the country. Full article

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

► Show Figures

Figure 1

18 pages, 2695 KiB

Open AccessArticle

Optimizing Patent Prior Art Search: An Approach Using Patent Abstract and Key Terms

by Amna Ali, Mohammad Ali Humayun, Liyanage Chandratilak De Silva and Pg Emeroylariffion Abas

Information 2025, 16(2), 145; https://doi.org/10.3390/info16020145 - 15 Feb 2025

Viewed by 750

Abstract

The rapid advancement of technology has led to a sustained accumulation of patent documents globally, as newly filed applications add to an ever-expanding repository of prior art. The need for innovation and progress within the patent system underscores the significance of robust patent [...] Read more.

The rapid advancement of technology has led to a sustained accumulation of patent documents globally, as newly filed applications add to an ever-expanding repository of prior art. The need for innovation and progress within the patent system underscores the significance of robust patent investigation, which includes prior art searches. The swift expansion of the patent arena poses challenges for experts employing conventional qualitative practices to handle the increasing quantitative needs. In this study, we propose a novel method to enhance patent prior art search through the integration of advanced natural language processing (NLP) techniques. Our approach leverages the abstract and top terms of patent documents to generate a unique set of labelled databases. This database is then utilized to train Bidirectional Encoder Representations from Transformers (BERT) for patents, enabling domain-specific prior art searches. Testing our method on the Google Public Patent Database yielded an improved F1 score of 0.94 on the testing data. Not only does our method demonstrate superior accuracy compared to baseline approaches, but it also exhibits enhanced computational efficiency. The refined prior art search promises to provide valuable assistance to specialists in their decision-making processes, offering insightful analyses and relevant information that can significantly increase the efficiency and accuracy of their judgments. Full article

► Show Figures

Graphical abstract

28 pages, 4157 KiB

Open AccessArticle

Integrating Quantitative Analyses of Historical and Contemporary Apparel with Educational Applications

by Zlatina Kazlacheva, Daniela Orozova, Nadezhda Angelova, Elena Zurleva, Julieta Ilieva and Zlatin Zlatev

Information 2025, 16(2), 144; https://doi.org/10.3390/info16020144 - 15 Feb 2025

Viewed by 584

Abstract

In this paper, a comparative analysis of historical and contemporary fashion designs was conducted using quantitative methods and indices. Elements such as silhouettes, color palettes, and structural characteristics were analyzed in order to identify models for reinterpretation of classic fashion costume. Clothing from [...] Read more.

In this paper, a comparative analysis of historical and contemporary fashion designs was conducted using quantitative methods and indices. Elements such as silhouettes, color palettes, and structural characteristics were analyzed in order to identify models for reinterpretation of classic fashion costume. Clothing from four historical periods was studied: Empire, Romanticism, the Victorian era, and Art Nouveau. An image processing algorithm was proposed, through which data on the shapes and colors of historical and contemporary clothing were obtained from digital color images. The most informative of the shape and color indices of contemporary and historical clothing were selected using the RReliefF, FSRNCA, and SFCPP methods. The feature vectors were reduced using the latent variable and t-SNE methods. The obtained data were used to group the clothing according to historical periods. Using Euclidean distances, the relationship between clothing by contemporary designers and the elements of the historical costume used by them was determined. These results were used to create an educational and methodological framework for practical training of students in the field of fashion design. The results of this work can help contemporary designers in interpreting and integrating elements of historical fashion into their collections, adapting them to the needs and preferences of consumers. Full article

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

► Show Figures

Figure 1

17 pages, 766 KiB

Open AccessArticle

Semi-Supervised Relation Extraction Corpus Construction and Models Creation for Under-Resourced Languages: A Use Case for Slovene

by Timotej Knez, Miha Štravs and Slavko Žitnik

Information 2025, 16(2), 143; https://doi.org/10.3390/info16020143 - 15 Feb 2025

Viewed by 363

Abstract

The goal of relation extraction is to recognize head and tail entities in a document and determine a relation between them. While a lot of progress was made in solving automated relation extraction in widely used languages such as English, the use of [...] Read more.

The goal of relation extraction is to recognize head and tail entities in a document and determine a relation between them. While a lot of progress was made in solving automated relation extraction in widely used languages such as English, the use of these methods for under-resourced languages and domains is limited due to the lack of training data. In this work, we present a pipeline using distant supervision for constructing a relation extraction corpus in an arbitrary language. The corpus construction combines Wikipedia documents in the target language with relations in the WikiData knowledge graph. We demonstrate the process by constructing a new corpus for relation extraction in the Slovene language. Our corpus captures 20 unique relation types. The final corpus contains 811,032 relations annotated in 244,437 sentences. We use the corpus to train models using three architectures and evaluate them on the task of Slovene relation extraction. We achieve comparable performance to approaches on English data. Full article

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

► Show Figures

Figure 1

20 pages, 8214 KiB

Open AccessArticle

Convolutional Neural Network Applications in Current Sensor Fault Classification Mechanisms in Permanent Magnet Synchronous Motor Drive Systems

by Kamila Jankowska and Mateusz Dybkowski

Information 2025, 16(2), 142; https://doi.org/10.3390/info16020142 - 14 Feb 2025

Viewed by 453

Abstract

In this article, the possibilities of Convolutional Neural Network applications to classify stator current sensor faults in a vector-controlled drive with a Permanent Magnet Synchronous Motor are described. It was assumed that three basic faults, consisting of signal loss from the current sensor, [...] Read more.

In this article, the possibilities of Convolutional Neural Network applications to classify stator current sensor faults in a vector-controlled drive with a Permanent Magnet Synchronous Motor are described. It was assumed that three basic faults, consisting of signal loss from the current sensor, measurement noise, and gain error, can be effectively classified by the Convolutional Neural Networks. This work presents the results obtained in experimental research on a 0.894-kilowatt Permanent Magnet Synchronous Motor. Fault classification is based on raw phase current signals transformed into matrices. Classification is carried out using two neural structures operating in parallel for phases A and B. This article includes a description of the process of selecting input matrices, developing classifiers, and the experimental results in offline classification obtained at the efficiency level of 99.2% and 98.3% for phases A and B, respectively. This research was carried out for various operating conditions of the drive system. Full article

(This article belongs to the Special Issue Machine Learning and Artificial Intelligence with Applications)

► Show Figures

Figure 1

22 pages, 3691 KiB

Open AccessArticle

G-TS-HRNN: Gaussian Takagi–Sugeno Hopfield Recurrent Neural Network

by Omar Bahou, Mohammed Roudani and Karim El Moutaouakil

Information 2025, 16(2), 141; https://doi.org/10.3390/info16020141 - 14 Feb 2025

Viewed by 433

Abstract

The Hopfield Recurrent Neural Network (HRNN) is a single-point descent metaheuristic that uses a single potential solution to explore the search space of optimization problems, whose constraints and objective function are aggregated into a typical energy function. The initial point is usually randomly [...] Read more.

The Hopfield Recurrent Neural Network (HRNN) is a single-point descent metaheuristic that uses a single potential solution to explore the search space of optimization problems, whose constraints and objective function are aggregated into a typical energy function. The initial point is usually randomly initialized, then moved by applying operators, characterizing the discrete dynamics of the HRNN, which modify its position or direction. Like all single-point metaheuristics, HRNN has certain drawbacks, such as being more likely to get stuck in local optima or miss global optima due to the use of a single point to explore the search space. Moreover, it is more sensitive to the initial point and operator, which can influence the quality and diversity of solutions. Moreover, it can have difficulty with dynamic or noisy environments, as it can lose track of the optimal region or be misled by random fluctuations. To overcome these shortcomings, this paper introduces a population-based fuzzy version of the HRNN, namely Gaussian Takagi–Sugeno Hopfield Recurrent Neural Network (G-TS-HRNN). For each neuron, the G-TS-HRNN associates an input fuzzy variable of d values, described by an appropriate Gaussian membership function that covers the universe of discourse. To build an instance of G-TS-HRNN(s) of size s, we generate s n-uplets of fuzzy values that present the premise of the Takagi–Sugeno system. The consequents are the differential equations governing the dynamics of the HRNN obtained by replacing each premise fuzzy value with the mean of different Gaussians. The steady points of all the rule premises are aggregated using the fuzzy center of gravity equation, considering the level of activity of each rule. G-TS-HRNN is used to solve the random optimization method based on the support vector model. Compared with HRNN, G-TS-HRNN performs better on well-known data sets. Full article

► Show Figures

Figure 1

14 pages, 2226 KiB

Open AccessArticle

The Correlation Between Self-Assessment and Observation in Driving Style Classification: An On-Road Case Study

by Muhammad Zainul Abidin Kamaludin, Juffrizal Karjanto, Noryani Muhammad, Nidzamuddin Md Yusof, Muhammad Zahir Hassan, Mohamad Zairi Baharom, Zulhaidi Mohd Jawi and Matthias Rauterberg

Information 2025, 16(2), 140; https://doi.org/10.3390/info16020140 - 14 Feb 2025

Viewed by 441

Abstract

A typical classification of driving style from a human driver is conducted via self-assessment, which begs the question of the possibility of bias from the respondents. Although some research has been carried out validating the questionnaire, no controlled studies have yet to be [...] Read more.

A typical classification of driving style from a human driver is conducted via self-assessment, which begs the question of the possibility of bias from the respondents. Although some research has been carried out validating the questionnaire, no controlled studies have yet to be reported to validate the Malaysian driving style. This study aimed to validate the Malaysian driver using the Multidimensional Driving Style Inventory (MDSI) with five-factor driving styles (careful, risky, angry, anxious, and dissociative) in on-road situations. Forty-one respondents completed the experiment on two designated routes recorded over 45 min of driving. A modest correlation existed between the MDSI and the score retrieved from the on-road observation assessment. The result showed a low-to-medium correlation collected from acceleration in longitudinal directions compared with correlation analysis utilizing the MDSI scale. Exploring such latent traits is essential for precisely classifying human driver styles without bias. Full article

(This article belongs to the Special Issue Test and Evaluation Methods for Human-Machine Interfaces of Automated Vehicles II)

► Show Figures

Graphical abstract

23 pages, 522 KiB

Open AccessArticle

ORUD-Detect: A Comprehensive Approach to Offensive Language Detection in Roman Urdu Using Hybrid Machine Learning–Deep Learning Models with Embedding Techniques

by Nisar Hussain, Amna Qasim, Gull Mehak, Olga Kolesnikova, Alexander Gelbukh and Grigori Sidorov

Information 2025, 16(2), 139; https://doi.org/10.3390/info16020139 - 13 Feb 2025

Viewed by 582

Abstract

With the rapid expansion of social media, detecting offensive language has become critically important for healthy online interactions. This poses a considerable challenge for low-resource languages such as Roman Urdu which are widely spoken on platforms like Facebook. In this paper, we perform [...] Read more.

With the rapid expansion of social media, detecting offensive language has become critically important for healthy online interactions. This poses a considerable challenge for low-resource languages such as Roman Urdu which are widely spoken on platforms like Facebook. In this paper, we perform a comprehensive study of offensive language detection models on Roman Urdu datasets using both Machine Learning (ML) and Deep Learning (DL) approaches. We present a dataset of 89,968 Facebook comments and extensive preprocessing techniques such as TF-IDF features, Word2Vec, and fastText embeddings to address linguistic idiosyncrasies and code-mixed aspects of Roman Urdu. Among the ML models, a linear kernel Support Vector Machine (SVM) model scored the best performance, with an F1 score of 94.76, followed by SVM models with radial and polynomial kernels. Even the use of BoW uni-gram features with naive Bayes produced competitive results, with an F1 score of 94.26. The DL models performed well, with Bi-LSTM returning an F1 score of 98.00 with Word2Vec embeddings and fastText-based Bi-RNN performing at 97.00, showcasing the inference of contextual embeddings and soft similarity. The CNN model also gave a good result, with an F1 score of 96.00. The CNN model also achieved an F1 score of 96.00. This study presents hybrid ML and DL approaches to improve offensive language detection approaches for low-resource languages. This research opens up new doors to providing safer online environments for widespread Roman Urdu users. Full article

(This article belongs to the Special Issue Application of Machine Learning in Data Science and Computational Intelligence)

► Show Figures

Figure 1

18 pages, 3624 KiB

Open AccessArticle

Entropy and Stability in Blockchain Consensus Dynamics

by Aristidis G. Anagnostakis and Euripidis Glavas

Information 2025, 16(2), 138; https://doi.org/10.3390/info16020138 - 13 Feb 2025

Viewed by 570

Abstract

Every Blockchain architecture relies upon two major pillars: (a) the hash-based, block-binding mechanism and (b) the consensus-achievement mechanism. While the entropic behavior of (a) has been extensively studied in literature over the past decades, the same does not hold for (b). In this [...] Read more.

Every Blockchain architecture relies upon two major pillars: (a) the hash-based, block-binding mechanism and (b) the consensus-achievement mechanism. While the entropic behavior of (a) has been extensively studied in literature over the past decades, the same does not hold for (b). In this work, we explore the entropic behavior of the fully distributed Blockchain consensus mechanisms. We quantify the impact of witnessing as a consensus-achievement process under the perspectives of Shannon information entropy and Lyapunov stability. We demonstrate that Blockchain consensus, expressed as the complement of the collective disagreement in a system, is a Lyapunov function of the number of witnesses W. The more the witnessing in a system, the less the entropy of the system becomes, and it converges to more stable states. We prove that the entropy decline is steepest for low values of W. A new metric for the efficiency of the consensus process based on the Shannon information entropy is introduced, laying the foundations for future studies on Blockchain-based systems optimization. Full article

► Show Figures

Figure 1

22 pages, 1060 KiB

Open AccessArticle

Behavioral Drivers of AI Adoption in Banking in a Semi-Mature Digital Economy: A TAM and UTAUT-2 Analysis of Stakeholder Perspectives

by Aristides Papathomas, George Konteos and Giorgos Avlogiaris

Information 2025, 16(2), 137; https://doi.org/10.3390/info16020137 - 13 Feb 2025

Cited by 1 | Viewed by 1398

Abstract

The transformative potential of artificial intelligence (AI) in banking is widely acknowledged, yet its practical adoption often faces resistance from users. This study investigates the factors influencing AI adoption behavior among various stakeholders in the Greek semi-mature systemic banking ecosystem, addressing a critical [...] Read more.

The transformative potential of artificial intelligence (AI) in banking is widely acknowledged, yet its practical adoption often faces resistance from users. This study investigates the factors influencing AI adoption behavior among various stakeholders in the Greek semi-mature systemic banking ecosystem, addressing a critical gap in the relevant research. By utilizing the Technology Acceptance Model (TAM), Unified Theory of Acceptance and Use of Technology 2 (UTAUT-2), and Partial Least Squares Structural Equation Modelling (PLS-SEM) models, data from 297 respondents (bank employees, digital professionals, and the general public) were analyzed. The results highlight the strong relevance of constructs such as Performance Expectancy, Effort Expectancy, and Hedonic Motivation, whereas Social Influence was deemed non-significant, reflecting a pragmatic stance toward AI. Demographic factors like gender and age were found to have no significant moderating effect, challenging traditional stereotypes. However, occupation and education emerged as significant moderators, indicating varying attitudes among professions and educational levels. This study is the first to develop a theoretical framework for AI adoption by Greek banking institutions, offering Greek banking practitioners actionable insights. The findings also hold relevance for countries with similar digital maturity levels, aiding broader AI integration in banking. Full article

(This article belongs to the Special Issue New Information Communication Technologies in the Digital Era)

► Show Figures

Graphical abstract

27 pages, 2910 KiB

Open AccessReview

A Survey on Multimodal Large Language Models in Radiology for Report Generation and Visual Question Answering

by Ziruo Yi, Ting Xiao and Mark V. Albert

Information 2025, 16(2), 136; https://doi.org/10.3390/info16020136 - 12 Feb 2025

Viewed by 1990

Abstract

Large language models (LLMs) and large vision models (LVMs) have driven significant advancements in natural language processing (NLP) and computer vision (CV), establishing a foundation for multimodal large language models (MLLMs) to integrate diverse data types in real-world applications. This survey explores the [...] Read more.

Large language models (LLMs) and large vision models (LVMs) have driven significant advancements in natural language processing (NLP) and computer vision (CV), establishing a foundation for multimodal large language models (MLLMs) to integrate diverse data types in real-world applications. This survey explores the evolution of MLLMs in radiology, focusing on radiology report generation (RRG) and radiology visual question answering (RVQA), where MLLMs leverage the combined capabilities of LLMs and LVMs to improve clinical efficiency. We begin by tracing the history of radiology and the development of MLLMs, followed by an overview of MLLM applications in RRG and RVQA, detailing core datasets, evaluation metrics, and leading MLLMs that demonstrate their potential in generating radiology reports and answering image-based questions. We then discuss the challenges MLLMs face in radiology, including dataset scarcity, data privacy and security, and issues within MLLMs such as bias, toxicity, hallucinations, catastrophic forgetting, and limitations in traditional evaluation metrics. Finally, this paper proposes future research directions to address these challenges, aiming to help AI researchers and radiologists overcome these obstacles and advance the study of MLLMs in radiology. Full article

(This article belongs to the Special Issue Natural Language Processing (NLP) with Applications and Natural Language Understanding (NLU))

► Show Figures

Graphical abstract

18 pages, 1573 KiB

Open AccessArticle

PD-Net: Parkinson’s Disease Detection Through Fusion of Two Spectral Features Using Attention-Based Hybrid Deep Neural Network

by Munira Islam, Khadija Akter, Md. Azad Hossain and M. Ali Akber Dewan

Information 2025, 16(2), 135; https://doi.org/10.3390/info16020135 - 12 Feb 2025

Viewed by 1101

Abstract

Parkinson’s disease (PD) is a progressive degenerative brain disease that worsens with age, causing areas of the brain to weaken. Vocal dysfunction often emerges as one of the earliest and most prominent indicators of Parkinson’s disease, with a significant number of patients exhibiting [...] Read more.

Parkinson’s disease (PD) is a progressive degenerative brain disease that worsens with age, causing areas of the brain to weaken. Vocal dysfunction often emerges as one of the earliest and most prominent indicators of Parkinson’s disease, with a significant number of patients exhibiting vocal impairments during the initial stages of the illness. In view of this, to facilitate the diagnosis of Parkinson’s disease through the analysis of these vocal characteristics, this study focuses on exerting a combination of mel spectrogram and MFCC as spectral features. This study adopts Italian raw audio data to establish an efficient detection framework specifically designed to classify the vocal data into two distinct categories: healthy individuals and patients diagnosed with Parkinson’s disease. To this end, the study proposes a hybrid model that integrates Convolutional Neural Networks (CNNs) and Long Short-Term Memory networks (LSTMs) for the detection of Parkinson’s disease. Certainly, CNNs are employed to extract spatial features from the extracted spectro-temporal characteristics of vocal data, while LSTMs capture temporal dependencies, accelerating a comprehensive analysis of the development of vocal patterns over time. Additionally, the merging of a multi-head attention mechanism significantly enhances the model’s ability to concentrate on essential details, hence improving its overall performance. This unified method aims to enhance the detection of subtle vocal changes associated with Parkinson’s, enhancing overall diagnostic accuracy. The findings declare that this model achieves a noteworthy accuracy of 99.00% for the Parkinson’s disease detection process. Full article

(This article belongs to the Special Issue Feature Papers in Information in 2024–2025)

► Show Figures

Graphical abstract

25 pages, 7982 KiB

Open AccessArticle

Aerial Imagery Redefined: Next-Generation Approach to Object Classification

by Eran Dahan, Itzhak Aviv and Tzvi Diskin

Information 2025, 16(2), 134; https://doi.org/10.3390/info16020134 - 11 Feb 2025

Viewed by 721

Abstract

Identifying and classifying objects in aerial images are two significant and complex issues in computer vision. The fine-grained classification of objects in overhead images has become widespread in various real-world applications, due to recent advancements in high-resolution satellite and airborne imaging systems. The [...] Read more.

Identifying and classifying objects in aerial images are two significant and complex issues in computer vision. The fine-grained classification of objects in overhead images has become widespread in various real-world applications, due to recent advancements in high-resolution satellite and airborne imaging systems. The task is challenging, particularly in low-resource cases, due to the minor differences between classes and the significant differences within each class caused by the fine-grained nature. We introduce Classification of Objects for Fine-Grained Analysis (COFGA), a recently developed dataset for accurately categorizing objects in high-resolution aerial images. The COFGA dataset comprises 2104 images and 14,256 annotated objects across 37 distinct labels. This dataset offers superior spatial information compared to other publicly available datasets. The MAFAT Challenge is a task that utilizes COFGA to improve fine-grained classification methods. The baseline model achieved a mAP of 0.6. This cost was 60, whereas the most superior model achieved a score of 0.6271 by utilizing state-of-the-art ensemble techniques and specific preprocessing techniques. We offer solutions to address the difficulties in analyzing aerial images, particularly when annotated and imbalanced class data are scarce. The findings provide valuable insights into the detailed categorization of objects and have practical applications in urban planning, environmental assessment, and agricultural management. We discuss the constraints and potential future endeavors, specifically emphasizing the potential to integrate supplementary modalities and contextual information into aerial imagery analysis. Full article

(This article belongs to the Special Issue Online Registration and Anomaly Detection of Cyber Security Events)

► Show Figures

Figure 1

21 pages, 1186 KiB

Open AccessArticle

blockHealthSecure: Integrating Blockchain and Cybersecurity in Post-Pandemic Healthcare Systems

by Bishwo Prakash Pokharel, Naresh Kshetri, Suresh Raj Sharma and Sobaraj Paudel

Information 2025, 16(2), 133; https://doi.org/10.3390/info16020133 - 11 Feb 2025

Viewed by 2458

Abstract

The COVID-19 pandemic exposed critical vulnerabilities in global healthcare systems, particularly in data security and interoperability. This paper introduces the blockHealthSecure Framework, which integrates blockchain technology with advanced cybersecurity measures to address these weaknesses and build resilient post-pandemic healthcare systems. Blockchain’s decentralized and [...] Read more.

The COVID-19 pandemic exposed critical vulnerabilities in global healthcare systems, particularly in data security and interoperability. This paper introduces the blockHealthSecure Framework, which integrates blockchain technology with advanced cybersecurity measures to address these weaknesses and build resilient post-pandemic healthcare systems. Blockchain’s decentralized and immutable architecture enhances the accuracy, transparency, and protection of electronic medical records (EMRs) and sensitive healthcare data. Additionally, it facilitates seamless and secure data sharing among healthcare providers, addressing long-standing interoperability challenges. This study explores the challenges and benefits of blockchain integration in healthcare, with a focus on regulatory and ethical considerations such as HIPAA and GDPR compliance. Key contributions include detailed case studies and examples that demonstrate blockchain’s ability to mitigate risks like ransomware, insider threats, and data breaches. This framework’s design leverages smart contracts, cryptographic hashing, and zero-trust architecture to ensure secure data management and proactive threat mitigation. The findings emphasize the framework’s potential to enhance data security, improve system adaptability, and support regulatory compliance in the face of evolving healthcare challenges. By bridging existing gaps in healthcare cybersecurity, the blockHealthSecure Framework offers a scalable, future-proof solution for safeguarding health outcomes and preparing for global health crises. Full article

► Show Figures

Graphical abstract

18 pages, 12629 KiB

Open AccessArticle

Leveraging AI-Generated Virtual Speakers to Enhance Multilingual E-Learning Experiences

by Sergio Miranda and Rosa Vegliante

Information 2025, 16(2), 132; https://doi.org/10.3390/info16020132 - 11 Feb 2025

Viewed by 752

Abstract

The growing demand for accessible and effective e-learning platforms has led to an increased focus on innovative solutions to address the challenges posed by the diverse linguistic backgrounds of learners. This paper explores the use of AI-generated virtual speakers to enhance multilingual e-learning [...] Read more.

The growing demand for accessible and effective e-learning platforms has led to an increased focus on innovative solutions to address the challenges posed by the diverse linguistic backgrounds of learners. This paper explores the use of AI-generated virtual speakers to enhance multilingual e-learning experiences. This study employs a system developed using Google Sheets and Google Script to create and manage multilingual courses, integrating AI-powered virtual speakers to deliver content in learners’ native languages. The e-learning platform used is a customized Moodle, and three courses were developed: “Mental Wellbeing in Mining”, “Rescue in the Mine”, and “Risk Assessment” for a European ERASMUS+ project. This study involved 147 participants from various educational and professional backgrounds. The main findings indicate that AI-generated virtual speakers significantly improve the accessibility of e-learning content. Participants preferred content in their native language and found AI-generated videos effective and engaging. This study concludes that AI-generated virtual speakers offer a promising approach to overcoming linguistic barriers in e-learning, providing personalized and adaptive learning experiences. Future research should focus on addressing ethical considerations, such as data privacy and algorithmic bias, and expanding the user base to include more languages and proficiency levels. Full article

(This article belongs to the Special Issue Advancing Educational Innovation with Artificial Intelligence)

► Show Figures

Figure 1

22 pages, 2064 KiB

Open AccessReview

Prescribing the Future: The Role of Artificial Intelligence in Pharmacy

by Hesham Allam

Information 2025, 16(2), 131; https://doi.org/10.3390/info16020131 - 11 Feb 2025

Cited by 1 | Viewed by 3068

Abstract

Integrating artificial intelligence (AI) into pharmacy operations and drug discovery represents a groundbreaking milestone in healthcare, offering unparalleled opportunities to revolutionize medication management, accelerate drug development, and deliver truly personalized patient care. This review examines the pivotal impact of AI in critical domains, [...] Read more.

Integrating artificial intelligence (AI) into pharmacy operations and drug discovery represents a groundbreaking milestone in healthcare, offering unparalleled opportunities to revolutionize medication management, accelerate drug development, and deliver truly personalized patient care. This review examines the pivotal impact of AI in critical domains, including drug discovery and development, drug repurposing, clinical trials, and pharmaceutical productivity enhancement. By significantly reducing human workload, improving precision, and shortening timelines, AI empowers the pharmaceutical industry to achieve ambitious objectives efficiently. This study delves into tools and methodologies enabling AI implementation, addressing ongoing challenges such as data privacy, algorithmic transparency, and ethical considerations while proposing actionable strategies to overcome these barriers. Furthermore, it offers insights into the future of AI in pharmacy, highlighting its potential to foster innovation, enhance efficiency, and improve patient outcomes. This research is grounded in a rigorous methodology, employing advanced data collection techniques. A comprehensive literature review was conducted using platforms such as PubMed, Semantic Scholar, and multidisciplinary databases, with AI-driven algorithms refining the retrieval of relevant and up-to-date studies. Systematic data scoping incorporated diverse perspectives from medical, pharmaceutical, and computer science domains, leveraging natural language processing for trend analysis and thematic content coding to identify patterns, challenges, and emerging applications. Modern visualization tools synthesized the findings into explicit graphical representations, offering a comprehensive view of the key role of AI in shaping the future of pharmacy and healthcare. Full article

(This article belongs to the Special Issue Artificial Intelligence and Data Science for Health)

► Show Figures

Graphical abstract

46 pages, 1971 KiB

Open AccessArticle

Text Classification: How Machine Learning Is Revolutionizing Text Categorization

by Hesham Allam, Lisa Makubvure, Benjamin Gyamfi, Kwadwo Nyarko Graham and Kehinde Akinwolere

Information 2025, 16(2), 130; https://doi.org/10.3390/info16020130 - 10 Feb 2025

Cited by 1 | Viewed by 2777

Abstract

The automated classification of texts into predefined categories has become increasingly prominent, driven by the exponential growth of digital documents and the demand for efficient organization. This paper serves as an in-depth survey of text classification and machine learning, consolidating diverse aspects of [...] Read more.

The automated classification of texts into predefined categories has become increasingly prominent, driven by the exponential growth of digital documents and the demand for efficient organization. This paper serves as an in-depth survey of text classification and machine learning, consolidating diverse aspects of the field into a single, comprehensive resource—a rarity in the current body of literature. Few studies have achieved such breadth, and this work aims to provide a unified perspective, offering a significant contribution to researchers and the academic community. The survey examines the evolution of machine learning in text categorization (TC), highlighting its transformative advantages over manual classification, such as enhanced accuracy, reduced labor, and adaptability across domains. It delves into various TC tasks and contrasts machine learning methodologies with knowledge engineering approaches, demonstrating the strengths and flexibility of data-driven techniques. Key applications of TC are explored, alongside an analysis of critical machine learning methods, including document representation techniques and dimensionality reduction strategies. Moreover, this study evaluates a range of text categorization models, identifies persistent challenges like class imbalance and overfitting, and investigates emerging trends shaping the future of the field. It discusses essential components such as document representation, classifier construction, and performance evaluation, offering a well-rounded understanding of the current state of TC. Importantly, this paper also provides clear research directions, emphasizing areas requiring further innovation, such as hybrid methodologies, explainable AI (XAI), and scalable approaches for low-resource languages. By bridging gaps in existing knowledge and suggesting actionable paths forward, this work positions itself as a vital resource for academics and industry practitioners, fostering deeper exploration and development in text classification. Full article

(This article belongs to the Section Information Applications)

► Show Figures

Graphical abstract

20 pages, 8021 KiB

Open AccessArticle

CNN 1D: A Robust Model for Human Pose Estimation

by Mercedes Hernández de la Cruz, Uriel Solache, Antonio Luna-Álvarez, Sergio Ricardo Zagal-Barrera, Daniela Aurora Morales López and Dante Mujica-Vargas

Information 2025, 16(2), 129; https://doi.org/10.3390/info16020129 - 10 Feb 2025

Viewed by 750

Abstract

The purpose of this research is to develop an efficient model for human pose estimation (HPE). The main limitations of the study include the small size of the dataset and confounds in the classification of certain poses, suggesting the need for more data [...] Read more.

The purpose of this research is to develop an efficient model for human pose estimation (HPE). The main limitations of the study include the small size of the dataset and confounds in the classification of certain poses, suggesting the need for more data to improve the robustness of the model in uncontrolled environments. The methodology used combines MediaPipe for the detection of key points in images with a CNN1D model that processes preprocessed feature sequences. The Yoga Poses dataset was used for the training and validation of the model, and resampling techniques, such as bootstrapping, were applied to improve accuracy and avoid overfitting in the training. The results show that the proposed model achieves 96% overall accuracy in the classification of five yoga poses, with accuracy metrics above 90% for all classes. The implementation of the CNN1D model instead of traditional 2D or 3D architectures accomplishes the goal of maintaining a low computational cost and efficient preprocessing of the images, allowing for its use on mobile devices and real-time environments. Full article

(This article belongs to the Section Artificial Intelligence)

► Show Figures

Figure 1

Journal Menu

Journal Browser

Information, Volume 16, Issue 2 (February 2025) – 88 articles

Further Information

Guidelines

MDPI Initiatives

Follow MDPI