Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Costel Marian Ionascu has a total of 16 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Decoding the Novelty Paradox: Predicting Breakthrough Innovations via AI-Driven Patent Network Analysis

    by Costel Marian IonașCu

    Understanding how knowledge diffuses within complex technology ecosystems is critical for anticipating long-term innovation impact and securing competitive advantage. However, traditional citation-based metrics suffer from severe right-censoring bias and fail to capture the latent semantic distance inherent in radical inventions. This study introduces a multi-window hybrid artificial intelligence framework to investigate the novelty paradox, the counterintuitive phenomenon where extremely novel inventions face delayed recognition and assimilation barriers. Analysing a massive corpus of 1.31 million European Patent Office patents over 5-, 7- and 10-year horizons, the research integrates advanced natural language processing with graph neural networks to evaluate both the semantic content and the citation network topology of patents. The empirical findings reveal three core insights. First, the predictive power of semantic indicators increases monotonically over time, capturing long-term technological trajectories with 3.8 times greater accuracy than short-term metrics. Second, the study provides robust causal evidence of a persistent novelty paradox, demonstrating a negative, concave relationship between extreme semantic novelty and technological impact, disproving the traditional inverted-U hypothesis. Third, graph-based topological architectures significantly outperform tabular machine learning models in long-term impact forecasting, proving that an invention's breakthrough status is deeply embedded in the relational topology of prior art. For innovation management, these results emphasize the necessity of adopting extended evaluation horizons (10+ years) to prevent the premature abandonment of radical innovations. Furthermore, the proposed AI-driven framework provides corporate R&D managers, venture capital investors and policymakers with a scalable, highly accurate tool for early-stage technology screening and strategic intellectual property portfolio management.

  • Open Access

    From Topology to Geometry: A Neural Ricci Flow Framework for Predicting Flash Crashes and Contagion

    by Abdul Kadar Muhammad Masum, Abul Kalam Azad, Chanda Rani Debi, Ramona Birǎu, Virgil Popescu, Costel Marian Ionascu

    Early warning of systemic financial instability is a crucial issue for regulators and asset managers, especially in detecting flash crashes and concealed contagion paths, which cannot be detected using conventional surveillance mechanisms. In this context, conventional surveillance mechanisms refer to linear correlation based monitoring, volatility threshold rules, and static network centrality measures that rely on fixed topological summaries. Traditional regulatory mechanisms are based on a set of static measures of topology, such as Degree Centrality, PageRank, and matrices of linear correlations that assess the size of assets in terms of positioning in a financial network, systematically ignoring small and highly leveraged institutions that are sometimes crucial bridges in financial networks.We propose a novel geometric deep learning architecture, the Neuro Ricci Flow, a geometric framework that combines Financial Returns and ESG Momentum into a single Riemannian manifold, which conceptualizes systemic risk as real world changes in market topology and not statistical volatility, as commonly interpreted. Financial returns and ESG momentum are selected as universally observable and high frequency market indicators that directly encode dynamic interactions and stress propagation in financial systems. The framework is designed to detect market level structural instability and contagion dynamics, rather than to directly infer macroeconomic phenomena such as currency debasement or monetary policy driven effects.We consider a Neural Ordinary Differential Equation to learn the underlying physics of Ricci Flow, with dynamical simulation of market manifold evolution to subsequently find hyperbolic singularities, in which an area of extreme negative curvature indicates structural rupture in the network. Empirical valuation shows better predictive power with 100 percent recall of the geometric risk indicator with five percent baseline risk. The deeper meaning further elaborates on the practical implications of this framework, which include accurate alarm mechanisms for central banks, assisted by Ricci Curvature Analysis and geometric techniques for immunizing the portfolios of asset managers in need of protection against contagion. We discover that geometric deep learning has provided a more sensitive paradigm of systemic risk models compared to traditional topological models.


  • Open Access

    Mapping Occupational Stress and Burnout in the Probation System: A Quantitative Approach

    by Cristina Ilie, Costel Marian Ionașcu, Andreea Mihaela Niță

    This study presents the first nationwide, system-level investigation of occupational stress and professional burnout among probation counselors in Romania, in the context of increasing caseloads, complex job demands and limited institutional support. Building on a comprehensive theoretical analysis, we employ a sociological research design involving a representative sample of 247 probation counselors from all 42 national probation services. Using the Maslach Burnout Inventory—Human Services Questionnaire, along with stress-related factors, we examine the prevalence, most important factors and typologies of burnout. Advanced quantitative techniques—including multiple linear regression, principal component analysis and K-means clustering—allow for a robust identification of key predictors of emotional exhaustion and three distinct psychosocial profiles: stress-resistant seniors, under involved younger staff and overworked, frustrated employees. We also conducted a confirmatory factor analysis (CFA) to check the validity of the MBI-HSS. This typology offers a novel conceptual framework for understanding professional burnout in probation, highlighting systemic vulnerabilities and distinct risk categories. Nevertheless, limitations exist: self-reported data may underestimate stress, and omitting variables like resilience or work meaning constrains explanatory depth. Despite these constraints, this study addresses a significant gap in Romanian probation research and lays the foundation for future longitudinal and qualitative studies. These should incorporate psychological and organizational factors to improve targeted interventions and human resources strategies.


  • Open Access

    Artificial Intelligence Adoption in the European Union: A Data-Driven Cluster Analysis (2021–2024)

    by Costel Marian Ionașcu

    The adoption of artificial intelligence by enterprises in the EU countries increased significantly between 2021 and 2024, but the recorded values were uneven and very small. This study analyzed the main characteristics of the artificial intelligence adoption process, its dynamics and patterns using principal component analysis and K-means clustering. The results highlighted a shift from using technologies for process automation to more advanced ones like natural language generation. The process was extended and gradually covered almost all business areas. The lack of relevant expertise, high costs and gaps in regulation of the development and use of artificial intelligence are the important barriers identified by 2024. The cluster analysis of EU countries highlighted the existence of two permanent clusters, one containing the leading countries and one containing the countries lagging behind, showing a large gap between them. The increasing dependence on externally developed solutions has characterized a maturing market for artificial intelligence. The equitable adoption of artificial intelligence at the level of EU countries must be based on specific workforce training, investments in infrastructure, financial incentives and, last but not least, on clear regulations. Only in this way can the gap in this area at the EU level be reduced.


  • Introduction to Data Analytics in e-Learning

    by C. M. Ionaşcu, P. S. Popescu, M. L. Mocanu, M. C. Mihăescu

  • Finding Patterns of Students from Activity Data

    by Dumitru Dan Burdescu, Marian Cristian Mihăescu, Costel Marian Ionascu, Bogdan Logofatu

  • e-Learning Recommender System based on Linear Regression

    by Dumitru Dan Burdescu, Marian Cristian Mihăescu, Costel Marian Ionascu, Bogdan Logofatu

  • Design of a Social Intelligent Learner for e-Learning Environments

    by Dumitru Dan Burdescu, Marian Cristian Mihascu, Bogdan Logofatu, Costel Marian Ionascu

    One of the most important things a learner misses within an e-Learning environment is the social aspect. This paper proposes a design for a social intelligent learner (SIL) that will act as an agent on behalf of the actual learner. The overall system works as a community support system designed for e-Learning environments. The actual learners benefit by using the obtained computer mediated communication system that uses extensively the performed activities in driving interpersonal communication.

  • Support system for e-Learning environment based on learning activities and processes

    by Dumitru Dan Burdescu, Marian Cristian Mihaescu, Costel Marian Ionascu, Bogdan Logofatu

    In e-Learning domain there has been observed an important effort towards designing and integrating support system software. These software systems may be fully integrated or run as services along e-Learning platforms and have as main goal increasing the effectiveness of the e-Learning processes. This paper proposes a novel structure of a support system for e-Learning infrastructure that is based on data representing learner's activities and processes. Data relevance is based on well discipline structuring based on concept maps. Markov chain modeling and classification is used as main intelligent procedure for data analysis.


  • Adaptation and Intelligent Retrieval of Learning Content to Enhance On-line Courses

    by Marian Cristian Mihaescu, Dumitru Dan Burdescu, Costel Marian Ionascu, Bogdan Logofatu

    Enhancing on-line courses is one of the most important issues of current e-Learning platforms. These platforms try to upgrade from simple environments where learners can download documents and take tests to complex environments where the learning content is adapted and intelligently retrieved. This paper presents a methodology for achieving such goals in an attempt to enhance on-line courses. The methodology makes intensive use of well structured content, embedded logging mechanism for learner's performed activities and usage of state-of-the-art algorithms for proper knowledge acquisition.