APPLIED INFORMATICS · AI RESEARCH · SECURE SYSTEMS

Anastasia
Antonopoulou

AI & NLP · Cybersecurity · Information Systems

MSc student in Applied Informatics at Harokopio University, with research interests in AI evaluation, responsible AI, cybersecurity and information systems.

AI DECISION SUPPORT LLM XAI MCDM DATA EVALUATION

Research output

Accepted for publication · IEEE · 2026

How Many Dimensions of Intelligence? A Data Driven Clustering of LLMaaS Benchmarks

Evangelia Filiopoulou, Georgia Dede, Anastasia Antonopoulou, George Dimitrakopoulos & Mara Nikolaidou

A data-driven investigation of benchmark redundancy across 423 Large Language Models and 14 intelligence benchmarks using Spearman correlation and hierarchical clustering.

↓ Download PDF

Accepted manuscript. Official IEEE publication link will be added when available.

Academic CV

Academic background, research experience, publications, technical skills and professional experience.

AI, evaluation & decision making.

I hold a BSc in Informatics and Telematics from Harokopio University of Athens and I am currently pursuing an MSc in Applied Informatics, specialising in Information Systems and Digital Transformation.

My undergraduate research examined the cost–intelligence trade-off in Large Language Models as a Service (LLMaaS) through a hybrid Multi-Criteria Decision Making framework.

How can we select AI systems not only by performance, but by what actually matters to the people and organisations using them?

My broader research interests include explainable and responsible AI, AI evaluation, decision-support systems, Large Language Models and the responsible adoption of artificial intelligence within organisations.

My current interests also extend to cybersecurity in healthcare information systems and the modeling of complex software-intensive systems. I am interested in how intelligent technologies can be evaluated, designed and adopted securely and responsibly.

Research interests

My work combines computational analysis, AI evaluation and decision-support methodologies to study how artificial intelligence systems can be selected and deployed more transparently.

LLM Evaluation

Evaluation of Large Language Models across intelligence, performance, operational cost and deployment criteria.

Decision Support

Multi-Criteria Decision Making approaches for complex AI and technology-selection problems.

Explainable AI

Exploring how AI recommendations can become understandable and actionable for real-world decision makers.

Responsible AI

Reliability, fairness, privacy, robustness and accountability in AI evaluation and selection.

AI Benchmarking

Investigating redundancy and structure among benchmarks used to evaluate modern Large Language Models.

Digital Transformation

Understanding how organisations can evaluate and adopt emerging AI technologies responsibly.

Healthcare Cybersecurity

Ongoing exploration of cybersecurity compliance, information security and the protection of sensitive healthcare information.

Systems Engineering

Requirements analysis, UML modeling and the design of interconnected information systems, including emerging educational technology applications.

Academic background

MSc in Applied Informatics

Harokopio University of Athens

Information Systems and Digital Transformation

Oct 2026 – Present

BSc in Informatics and Telematics

Harokopio University of Athens

Information Systems Management · Final Grade: 7.14 / 10

2022 – 2026

Research work

Assessing the Cost–Intelligence Trade-off in Large Language Models via CRITIC–BWM

Harokopio University of Athens · 2026

Developed a hybrid Multi-Criteria Decision Making framework for evaluating commercially available Large Language Models by balancing intelligence performance with operational cost.

The methodology combined hierarchical clustering, CRITIC objective weighting and Best-Worst Method subjective weighting based on assessments from 19 industry experts. VIKOR was used for compromise ranking, followed by sensitivity analysis to evaluate ranking robustness.

Hierarchical Clustering · CRITIC · BWM · VIKOR · Sensitivity Analysis · LLM Evaluation

Cybersecurity Compliance in Health Issue

Independent research · October 2026 – Present · Ongoing

Newly initiated research exploring cybersecurity compliance in healthcare contexts, with a focus on the intersection of information security, regulatory requirements and the protection of sensitive health information.

Cybersecurity · Compliance · Healthcare · Information Security

Research & technical skills

Research Methods

Hierarchical Clustering
Spearman Correlation
CRITIC
Best-Worst Method (BWM)
VIKOR
Sensitivity Analysis
Multi-Criteria Decision Making

Programming & Technologies

Python
Streamlit
Spring Boot
REST APIs
Git & GitHub
Web Development

Get in touch.

For research collaborations, academic opportunities or professional enquiries, feel free to reach out.

Document
↓ Download PDF