Computational and Mathematical Methods in Information Science and Engineering, 2nd Edition

This Special Issue compiles eleven excellent papers, selected from a highly competitive pool of submissions, to collectively highlight the dynamic and interdisciplinary nature of the crucial field of computational and mathematical methods in information science and engineering. Published between August 2023 and May 2025, these contributions represent significant steps forward in leveraging advanced computational and mathematical techniques to address complex problems across diverse domains.

1. Recent Developments and Knowledge Gaps

The past few years have witnessed exponential growth in the volume and complexity of data, driven by advancements in digital technologies, sensor networks, and interconnected systems. This proliferation has underscored the urgent need for sophisticated computational and mathematical methods to extract meaningful insights, optimize processes, and ensure the robustness of information systems, as foundational, traditional approaches often struggle with the scale, high dimensionality, and inherent noise present in real-world data. Key persistent challenges include the following:

• Handling Heterogeneity and Uncertainty: Real-world systems—from carbon emission analysis to customer behavior modeling—exhibit complex heterogeneities and uncertainties that defy simplistic modeling. Developing robust frameworks that can accurately capture, quantify, and decompose these variabilities remains a significant challenge.
• Optimizing Complex, Interconnected Systems: Whether in microgrid design, supply chain management, or urban planning, optimizing numerous interdependent components for multiple, often conflicting, objectives is critical. This necessitates novel algorithms and computational paradigms capable of navigating vast, complex solution spaces efficiently.
• Extracting Actionable Insights from Unstructured Data: The immense volume of unstructured data, such as online reviews and sensor readings, presents both opportunities and challenges. Bridging the gap between raw data and actionable knowledge requires advanced analytical techniques, particularly in areas like fine-grained sentiment analysis and human activity recognition.
• Ensuring Security and Privacy in Pervasive Computing: As smart devices become ubiquitous, ensuring the security and privacy of user data, especially through implicit and continuous authentication methods, is paramount. Developing robust and accurate authentication systems that perform reliably in noisy, real-world environments remains an active research frontier.
• Modeling Dynamic and Interconnected Systems: Modern systems are rarely static; they evolve over time and exhibit intricate interdependencies. Capturing these dynamics, whether in innovation networks or social interactions, requires sophisticated models that go beyond static analyses to account for temporal evolution and complex network effects.

2. How This Special Issue Addresses the Gaps

This Special Issue makes substantial contributions toward addressing these critical knowledge gaps, showcasing innovative applications of computational and mathematical methods. Several papers demonstrate sophisticated, nuanced approaches to efficiency and data analysis. Contribution 1 introduces a novel meta-frontier framework integrated with machine-learning-driven projections to accurately analyze carbon emission efficiency, effectively resolving inconsistencies inherent in traditional models and offering refined policy implications. Similarly, Contribution 2 employs non-radial directional distance functions and meta-frontier models to evaluate China’s high-quality development efficiency, providing new insights into the role of higher education labor. The power of machine learning to extract deeper, more actionable insights is a central theme of this contribution. Contribution 3 utilizes multiple correspondence analysis combined with hierarchical clustering and various machine learning classifiers (GBM, RF, SVM) to model e-customer profiles, yielding valuable insights for understanding consumer behavior. Contribution 4 effectively demonstrates the use of an XGBoost model, optimized via a Cuckoo search algorithm, to achieve high accuracy in caregiver activity recognition using sensor data, thereby addressing a vital need in elderly care. Furthermore, Contribution 5 proposes a practical two-tier optimization model for rural microgrids that balances enterprise costs with user satisfaction, offering actionable configuration strategies for sustainable energy solutions. Understanding complex network dynamics is another recurring and vital theme of this paper. Contribution 6 introduces an innovative double-layer coupled network model to analyze innovation efficiency, revealing the nuanced impact of online and offline network densities. Contribution 7 applies exponential random graph models (ERGMs) to characterize multiplex health social networks, providing a deeper, more structured understanding of patient engagement within online health communities. The predictive capabilities of mathematical and computational techniques are also significantly enhanced. Contribution 8 combines wavelet denoising with long short-term memory (LSTM) networks to achieve highly accurate short-term rail transit passenger flow prediction, offering practical solutions for urban transportation management. Contribution 9 proposes a multi-stage ECRM model that integrates fine-grained sentiment analysis with stochastic dominance criteria to accurately evaluate the perceived quality of clothing brands based on online user reviews. The critical societal impact of digital advancements is thoughtfully explored. Contribution 10 investigates the positive effects of digital finance on green low-carbon circular development using machine learning models, identifying technological innovation as a key mediating factor. Finally, Contribution 11 addresses pressing security concerns by presenting ST-SVD—a real-time, sensor-based mobile user authentication approach that utilizes a semi-supervised teacher–student tri-training algorithm with CNN, achieving notably high authentication accuracy.

3. Future Research Directions

While this Special Issue showcases remarkable progress in the field of computational and mathematical methods in information science and engineering, it continues to evolve rapidly, presenting exciting and critical avenues for future research:

• Explainable AI (XAI) and Trustworthy AI: As complex models like deep neural networks become more prevalent, understanding their internal decision-making processes is crucial, especially in high-stakes applications like healthcare and finance. Future research must focus on developing robust XAI techniques that provide transparency and interpretability without sacrificing performance.
• Federated Learning and Privacy-Preserving AI: Amid growing global concerns about data privacy, federated learning and other privacy-preserving AI techniques are of paramount importance. Research into efficient, secure, and scalable methods for collaborative model training on decentralized, sensitive datasets is essential.
• Quantum Computing for Information Science: The emerging field of quantum computing holds immense potential for solving classically intractable optimization problems and performing advanced data analysis. Exploring the application of quantum algorithms to areas like cryptography, large-scale data processing, and complex system modeling represents a significant future frontier.
• Integration of Multi-Modal Data: Real-world information is inherently multi-modal (including text, image, audio, and sensor data). Future research should focus on developing unified computational and mathematical frameworks that can effectively integrate, synchronize, and analyze these diverse data types for more holistic and comprehensive insights.
• Ethical AI and Algorithmic Fairness: As AI systems become deeply integrated into societal structures, addressing ethical considerations, bias, and ensuring fairness in algorithms is critical. Research is needed to develop rigorous methods for detecting, mitigating, and preventing algorithmic bias to ensure equitable and just outcomes.
• Causality in Machine Learning: Moving beyond correlation to establish causation is a fundamental challenge. Future work should explore integrating causal inference methods with machine learning to build models that not only predict but also explain underlying causal relationships, leading to more robust and reliable interventions.
• Dynamic and Adaptive Systems: Developing computational and mathematical models that can continuously learn and adapt to non-stationary, changing environments—whether in real-time sensor data analysis or evolving social networks—will be crucial for building truly intelligent and resilient systems. This includes advanced research into reinforcement learning in complex, real-world scenarios.
• Computational Social Science and Digital Humanities: Applying these powerful methods to analyze vast amounts of social and cultural data offers unprecedented opportunities to understand human behavior, societal trends, and historical patterns. This richly interdisciplinary area holds significant promise for the future.