Nikolaos Ersotelos
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Nikolaos Ersotelos has a total of 55 co-authors. Here are the top co-authors by number of shared publications:
Feng Dong
University of Bedfordshire, Luton, UK
12 shared publicationsEnjie Liu
University of Bedfordshire
5 shared publicationsFarzad Parvinzamir
InsightZen LLC, Hangzhou, China
5 shared publicationsYoubing Zhao
University of Bedfordshire, Luton, UK
5 shared publications- Gregory Epiphaniou
University of Warwick, Coventry, United Kingdom
4 shared publications Xia Zhao
University of Bedfordshire
4 shared publications
Latest Publications
- Open Access
- 14 February 2026
- Published by MDPI AG in Future Internet
- Vol. 18 (2) , 101
A Temporally Dynamic Feature-Extraction Framework for Phishing Detection with LIME and SHAP Explanations
Phishing remains one of the most pervasive social engineering threats, exploiting human vulnerabilities and continuously evolving to bypass static detection mechanisms. Existing machine learning models achieve high accuracy but often act as opaque systems that lack robustness to evolving tactics and explainability, limiting trust and real-world deployment. In this research, we propose a dynamic Explainable AI (XAI) approach for phishing detection that integrates temporally aware feature extraction with dual interpretability through LIME and SHAP applied to the resulting window-level features. The novelty of this research lies in a temporally dynamic feature framework that simulates a plausible email reading progression using a heuristic temporal model and employs a sliding window aggregation method to capture behavioural and temporal patterns within email content. Using an aggregated dataset of 82,500 phishing and legitimate emails, dynamic features were extracted and used to train four classifiers: Random Forest, XGBoost, Multi-Layer Perceptron, and Logistic Regression. Ensemble models demonstrated strong performance with XGBoost achieving 94% accuracy and Random Forest 93%. This research addresses an important gap by combining dynamically constructed temporal features with transparent explanations, achieving high detection performance while preserving interpretability. These findings demonstrate that dynamic temporal modelling with explainable learning can enhance the trustworthiness and practicality of phishing detection systems, highlighting that temporally structured features and explainable learning can enhance the trustworthiness and practical deployability of phishing detection systems without incurring excessive computational overhead.
- 14 February 2026
- Published by MDPI AG in Future Internet
- Vol. 18 (2) , 101
Open AccessA Temporally Dynamic Feature-Extraction Framework for Phishing Detection with LIME and SHAP Explanations
Phishing remains one of the most pervasive social engineering threats, exploiting human vulnerabilities and continuously evolving to bypass static detection mechanisms. Existing machine learning models achieve high accuracy but often act as opaque systems that lack robustness to evolving tactics and explainability, limiting trust and real-world deployment. In this research, we propose a dynamic Explainable AI (XAI) approach for phishing detection that integrates temporally aware feature extraction with dual interpretability through LIME and SHAP applied to the resulting window-level features. The novelty of this research lies in a temporally dynamic feature framework that simulates a plausible email reading progression using a heuristic temporal model and employs a sliding window aggregation method to capture behavioural and temporal patterns within email content. Using an aggregated dataset of 82,500 phishing and legitimate emails, dynamic features were extracted and used to train four classifiers: Random Forest, XGBoost, Multi-Layer Perceptron, and Logistic Regression. Ensemble models demonstrated strong performance with XGBoost achieving 94% accuracy and Random Forest 93%. This research addresses an important gap by combining dynamically constructed temporal features with transparent explanations, achieving high detection performance while preserving interpretability. These findings demonstrate that dynamic temporal modelling with explainable learning can enhance the trustworthiness and practicality of phishing detection systems, highlighting that temporally structured features and explainable learning can enhance the trustworthiness and practical deployability of phishing detection systems without incurring excessive computational overhead.
- Open Access
- 1 January 2026
- Published by Elsevier in Vehicular Communications
- Vol. 58
Challenges and opportunities of synthetic data generation for machine learning-based intrusion detection systems in in-vehicle networks
- 1 January 2026
- Published by Elsevier in Vehicular Communications
- Vol. 58
Open AccessChallenges and opportunities of synthetic data generation for machine learning-based intrusion detection systems in in-vehicle networks
- 8 September 2025
- Published by Elsevier in Journal of Space Safety Engineering
- Vol. 12 (4) , 730-748
Operationalisation of industry best practices in SSE standardisation for space mission systems
- 8 September 2025
- Published by Elsevier in Journal of Space Safety Engineering
- Vol. 12 (4) , 730-748
Operationalisation of industry best practices in SSE standardisation for space mission systems
- Open Access
A Hybrid Methodology to Assess Cyber Resilience of IoT in Energy Management and Connected Sites
Cyber threats and vulnerabilities present an increasing risk to the safe and frictionless execution of business operations. Bad actors (“hackers”), including state actors, are increasingly targeting the operational technologies (OTs) and industrial control systems (ICSs) used to protect critical national infrastructure (CNI). Minimisations of cyber risk, attack surfaces, data immutability, and interoperability of IoT are some of the main challenges of today’s CNI. Cyber security risk assessment is one of the basic and most important activities to identify and quantify cyber security threats and vulnerabilities. This research presents a novel i-TRACE security-by-design CNI methodology that encompasses CNI key performance indicators (KPIs) and metrics to combat the growing vicarious nature of remote, well-planned, and well-executed cyber-attacks against CNI, as recently exemplified in the current Ukraine conflict (2014–present) on both sides. The proposed methodology offers a hybrid method that specifically identifies the steps required (typically undertaken by those responsible for detecting, deterring, and disrupting cyber attacks on CNI). Furthermore, we present a novel, advanced, and resilient approach that leverages digital twins and distributed ledger technologies for our chosen i-TRACE use cases of energy management and connected sites. The key steps required to achieve the desired level of interoperability and immutability of data are identified, thereby reducing the risk of CNI-specific cyber attacks and minimising the attack vectors and surfaces. Hence, this research aims to provide an extra level of safety for CNI and OT human operatives, i.e., those tasked with and responsible for detecting, deterring, disrupting, and mitigating these cyber-attacks. Our evaluations and comparisons clearly demonstrate that i-TRACE has significant intrinsic advantages compared to existing “state-of-the-art” mechanisms.
Open AccessA Hybrid Methodology to Assess Cyber Resilience of IoT in Energy Management and Connected Sites
Cyber threats and vulnerabilities present an increasing risk to the safe and frictionless execution of business operations. Bad actors (“hackers”), including state actors, are increasingly targeting the operational technologies (OTs) and industrial control systems (ICSs) used to protect critical national infrastructure (CNI). Minimisations of cyber risk, attack surfaces, data immutability, and interoperability of IoT are some of the main challenges of today’s CNI. Cyber security risk assessment is one of the basic and most important activities to identify and quantify cyber security threats and vulnerabilities. This research presents a novel i-TRACE security-by-design CNI methodology that encompasses CNI key performance indicators (KPIs) and metrics to combat the growing vicarious nature of remote, well-planned, and well-executed cyber-attacks against CNI, as recently exemplified in the current Ukraine conflict (2014–present) on both sides. The proposed methodology offers a hybrid method that specifically identifies the steps required (typically undertaken by those responsible for detecting, deterring, and disrupting cyber attacks on CNI). Furthermore, we present a novel, advanced, and resilient approach that leverages digital twins and distributed ledger technologies for our chosen i-TRACE use cases of energy management and connected sites. The key steps required to achieve the desired level of interoperability and immutability of data are identified, thereby reducing the risk of CNI-specific cyber attacks and minimising the attack vectors and surfaces. Hence, this research aims to provide an extra level of safety for CNI and OT human operatives, i.e., those tasked with and responsible for detecting, deterring, disrupting, and mitigating these cyber-attacks. Our evaluations and comparisons clearly demonstrate that i-TRACE has significant intrinsic advantages compared to existing “state-of-the-art” mechanisms.
- Open Access
- 16 December 2022
- Published by MDPI AG in Applied Sciences
- Vol. 12 (24) , 12941
Reducing False Negatives in Ransomware Detection: A Critical Evaluation of Machine Learning Algorithms
Technological achievement and cybercriminal methodology are two parallel growing paths; protocols such as Tor and i2p (designed to offer confidentiality and anonymity) are being utilised to run ransomware companies operating under a Ransomware as a Service (RaaS) model. RaaS enables criminals with a limited technical ability to launch ransomware attacks. Several recent high-profile cases, such as the Colonial Pipeline attack and JBS Foods, involved forcing companies to pay enormous amounts of ransom money, indicating the difficulty for organisations of recovering from these attacks using traditional means, such as restoring backup systems. Hence, this is the benefit of intelligent early ransomware detection and eradication. This study offers a critical review of the literature on how we can use state-of-the-art machine learning (ML) models to detect ransomware. However, the results uncovered a tendency of previous works to report precision while overlooking the importance of other values in the confusion matrices, such as false negatives. Therefore, we also contribute a critical evaluation of ML models using a dataset of 730 malware and 735 benign samples to evaluate their suitability to mitigate ransomware at different stages of a detection system architecture and what that means in terms of cost. For example, the results have shown that an Artificial Neural Network (ANN) model will be the most suitable as it achieves the highest precision of 98.65%, a Youden’s index of 0.94, and a net benefit of 76.27%, however, the Random Forest model (lower precision of 92.73%) offered the benefit of having the lowest false-negative rate (0.00%). The risk of a false negative in this type of system is comparable to the unpredictable but typically large cost of ransomware infection, in comparison with the more predictable cost of the resources needed to filter false positives.
- 16 December 2022
- Published by MDPI AG in Applied Sciences
- Vol. 12 (24) , 12941
Open AccessReducing False Negatives in Ransomware Detection: A Critical Evaluation of Machine Learning Algorithms
Technological achievement and cybercriminal methodology are two parallel growing paths; protocols such as Tor and i2p (designed to offer confidentiality and anonymity) are being utilised to run ransomware companies operating under a Ransomware as a Service (RaaS) model. RaaS enables criminals with a limited technical ability to launch ransomware attacks. Several recent high-profile cases, such as the Colonial Pipeline attack and JBS Foods, involved forcing companies to pay enormous amounts of ransom money, indicating the difficulty for organisations of recovering from these attacks using traditional means, such as restoring backup systems. Hence, this is the benefit of intelligent early ransomware detection and eradication. This study offers a critical review of the literature on how we can use state-of-the-art machine learning (ML) models to detect ransomware. However, the results uncovered a tendency of previous works to report precision while overlooking the importance of other values in the confusion matrices, such as false negatives. Therefore, we also contribute a critical evaluation of ML models using a dataset of 730 malware and 735 benign samples to evaluate their suitability to mitigate ransomware at different stages of a detection system architecture and what that means in terms of cost. For example, the results have shown that an Artificial Neural Network (ANN) model will be the most suitable as it achieves the highest precision of 98.65%, a Youden’s index of 0.94, and a net benefit of 76.27%, however, the Random Forest model (lower precision of 92.73%) offered the benefit of having the lowest false-negative rate (0.00%). The risk of a false negative in this type of system is comparable to the unpredictable but typically large cost of ransomware infection, in comparison with the more predictable cost of the resources needed to filter false positives.
- Open Access
- 29 December 2020
- Published by MDPI AG in Journal of Sensor and Actuator Networks
- Vol. 10 (1) , 1
Blockchain and IoMT against Physical Abuse: Bullying in Schools as a Case Study
By law, schools are required to protect the well-being of students against problems such as on-campus bullying and physical abuse. In the UK, a report by the Office for Education (OfE) showed 17% of young people had been bullied during 2017–2018. This problem continues to prevail with consequences including depression, anxiety, suicidal thoughts, and eating disorders. Additionally, recent evidence suggests this type of victimisation could intensify existing health complications. This study investigates the opportunities provided by Internet of Medical Things (IoMT) data towards next-generation safeguarding. A new model is developed based on blockchain technology to enable real-time intervention triggered by IoMT data that can be used to detect stressful events, e.g., when bullying takes place. The model utilises private permissioned blockchain to manage IoMT data to achieve quicker and better decision-making while revolutionising aspects related to compliance, double-entry, confidentiality, and privacy. The feasibility of the model and the interaction between the sensors and the blockchain was simulated. To facilitate a close approximation of an actual IoMT environment, we clustered and decomposed existing medical sensors to their attributes, including their function, for a variety of scenarios. Then, we demonstrated the performance and capabilities of the emulator under different loads of sensor-generated data. We argue to the suitability of this emulator for schools and medical centres to conduct feasibility studies to address sensor data with disruptive data processing and management technologies.
- 29 December 2020
- Published by MDPI AG in Journal of Sensor and Actuator Networks
- Vol. 10 (1) , 1
Open AccessBlockchain and IoMT against Physical Abuse: Bullying in Schools as a Case Study
By law, schools are required to protect the well-being of students against problems such as on-campus bullying and physical abuse. In the UK, a report by the Office for Education (OfE) showed 17% of young people had been bullied during 2017–2018. This problem continues to prevail with consequences including depression, anxiety, suicidal thoughts, and eating disorders. Additionally, recent evidence suggests this type of victimisation could intensify existing health complications. This study investigates the opportunities provided by Internet of Medical Things (IoMT) data towards next-generation safeguarding. A new model is developed based on blockchain technology to enable real-time intervention triggered by IoMT data that can be used to detect stressful events, e.g., when bullying takes place. The model utilises private permissioned blockchain to manage IoMT data to achieve quicker and better decision-making while revolutionising aspects related to compliance, double-entry, confidentiality, and privacy. The feasibility of the model and the interaction between the sensors and the blockchain was simulated. To facilitate a close approximation of an actual IoMT environment, we clustered and decomposed existing medical sensors to their attributes, including their function, for a variety of scenarios. Then, we demonstrated the performance and capabilities of the emulator under different loads of sensor-generated data. We argue to the suitability of this emulator for schools and medical centres to conduct feasibility studies to address sensor data with disruptive data processing and management technologies.
- Open Access
- 2 August 2019
- Published by Springer Nature in The Visual Computer
- Vol. 36 (7) , 1337-1354
Literature Explorer: effective retrieval of scientific documents through nonparametric thematic topic detection
- 2 August 2019
- Published by Springer Nature in The Visual Computer
- Vol. 36 (7) , 1337-1354
Open AccessLiterature Explorer: effective retrieval of scientific documents through nonparametric thematic topic detection
- Open Access
- 13 October 2018
- Published by Springer Nature in Hormones
- Vol. 17 (4) , 541-550
Review of mobile applications for optimizing the follow-up care of patients with diabetes
Several smartphone applications aim at facilitating communication between patients and healthcare providers. In this review, we evaluate and compare the most promising applications in the field of diabetes mellitus (DM) and obesity. Most applications monitor body weight, fasting or postprandial blood glucose, glycosylated hemoglobin (Hgb) A1c (HgbA1c), and units and types of insulin used.
- 13 October 2018
- Published by Springer Nature in Hormones
- Vol. 17 (4) , 541-550
Open AccessReview of mobile applications for optimizing the follow-up care of patients with diabetes
Several smartphone applications aim at facilitating communication between patients and healthcare providers. In this review, we evaluate and compare the most promising applications in the field of diabetes mellitus (DM) and obesity. Most applications monitor body weight, fasting or postprandial blood glucose, glycosylated hemoglobin (Hgb) A1c (HgbA1c), and units and types of insulin used.
- Open Access
- 7 February 2017
- Published by Science Impact, Ltd. in Impact
- Vol. 2017 (2) , 40-44
Dr. Inventor, Promoting Scientific Creativity by Utilising Web-based Research Objects, FP7
Scientific creativity and innovation represent the beating heart of European growth at a time of rapid technological change. Dr Inventor is built on the vision that technologies have great potential to supplement human ingenuity in science by overcoming the limitations that people suffer in pursuing scientific discovery. It presents an original system that will provide inspiration for scientific creativity by utilising the rich presence of web-based research resources. Dr Inventor will act as a personal research assistant, utilising machine-empowered search and computation to bring researchers extended perspectives for scientific innovation by informing them of a broad spectrum of relevant research concepts and approaches, by assessing the novelty of research ideas, and by offering suggestions of new concepts and workflows with unexpected features for new scientific discovery. Dr Inventor is an attempt to understand the potential of technology in the scientific creative process within current technology limitations. It represents a sound balance between scientific insight into individual scientific creative processes and technical implementation using innovative technologies in information extraction, document summarization, semantics and visual analytics. The outcomes will be integrated into a web-based system that will allow evaluation in a selected research area under real-world settings with carefully designed metrics, benchmarks and baseline for creative performance, leading to tangible measurements on the performance of the technologies in enhancing human creativity and a blueprint for future technologies in computational creativity. Dr Inventor has huge implications for scientific innovation in Europe, as it has the potential to change the way in which scientific research is undertaken. The acceptance of the system by general research communities will open opportunities for many industrial sectors, leading to reinforced leadership of European industry.
- 7 February 2017
- Published by Science Impact, Ltd. in Impact
- Vol. 2017 (2) , 40-44
Open AccessDr. Inventor, Promoting Scientific Creativity by Utilising Web-based Research Objects, FP7
Scientific creativity and innovation represent the beating heart of European growth at a time of rapid technological change. Dr Inventor is built on the vision that technologies have great potential to supplement human ingenuity in science by overcoming the limitations that people suffer in pursuing scientific discovery. It presents an original system that will provide inspiration for scientific creativity by utilising the rich presence of web-based research resources. Dr Inventor will act as a personal research assistant, utilising machine-empowered search and computation to bring researchers extended perspectives for scientific innovation by informing them of a broad spectrum of relevant research concepts and approaches, by assessing the novelty of research ideas, and by offering suggestions of new concepts and workflows with unexpected features for new scientific discovery. Dr Inventor is an attempt to understand the potential of technology in the scientific creative process within current technology limitations. It represents a sound balance between scientific insight into individual scientific creative processes and technical implementation using innovative technologies in information extraction, document summarization, semantics and visual analytics. The outcomes will be integrated into a web-based system that will allow evaluation in a selected research area under real-world settings with carefully designed metrics, benchmarks and baseline for creative performance, leading to tangible measurements on the performance of the technologies in enhancing human creativity and a blueprint for future technologies in computational creativity. Dr Inventor has huge implications for scientific innovation in Europe, as it has the potential to change the way in which scientific research is undertaken. The acceptance of the system by general research communities will open opportunities for many industrial sectors, leading to reinforced leadership of European industry.