Paul Stefan Popescu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Paul Stefan Popescu has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
- Marian Cristian Mihăescu
University of Craiova
11 shared publications M. C. Mihăescu
University of Craiova
3 shared publicationsM. L. Mocanu
University of Craiova
3 shared publications- Mihai Lucian Mocanu
Faculty of Automatics, Computer Science and Electronics, University of Craiova, 200440 Craiova, Romania
2 shared publications - Oana Maria Teodorescu
University of Craiova,Faculty of Automatics, Computers and Electronics,Craiova,Romania
2 shared publications Andrei Balan
University of Craiova
1 shared publication
Latest Publications
- 30 August 2026
- Published by Scientia Socialis Ltd in Problems of Education in the 21st Century
- Vol. 84 (4)
MAIN DRIVERS OF CHATGPT ADOPTION BY ROMANIAN COMPUTER SCIENCE STUDENTS
The use of artificial intelligence technologies in education has huge potential to improve teaching and learning by providing personalised learning experiences, on-demand academic support, and automated feedback. The emergence of generative artificial intelligence has introduced a new category of educational technology that is changing how students search for information, solve problems, and engage with course materials. This transformation is especially relevant for computer science students, who can benefit from using AI tools not only for learning but also for programming. The purpose of this research was to investigate the main drivers influencing the adoption of the ChatGPT tool among Romanian students enrolled in a computer science educational program. An extended technology acceptance model (TAM) that includes output quality and perceived impact as external variables has been specified and tested with a sample of 165 university students. The results show that perceived enjoyment and perceived usefulness have the strongest direct influence on the behavioural intention to use ChatGPT. The influence of perceived ease of use on perceived usefulness and behavioural intention was not significant, suggesting that it is not a relevant variable for computer science students. Another finding is that taking into consideration indirect effects, perceived impact of AI tools on education and future professions, and perceived enjoyment emerge as equally important for the adoption of ChatGPT. Keywords: ChatGPT, computer science, university students, TAM, AI tools
- 30 August 2026
- Published by Scientia Socialis Ltd in Problems of Education in the 21st Century
- Vol. 84 (4)
MAIN DRIVERS OF CHATGPT ADOPTION BY ROMANIAN COMPUTER SCIENCE STUDENTS
The use of artificial intelligence technologies in education has huge potential to improve teaching and learning by providing personalised learning experiences, on-demand academic support, and automated feedback. The emergence of generative artificial intelligence has introduced a new category of educational technology that is changing how students search for information, solve problems, and engage with course materials. This transformation is especially relevant for computer science students, who can benefit from using AI tools not only for learning but also for programming. The purpose of this research was to investigate the main drivers influencing the adoption of the ChatGPT tool among Romanian students enrolled in a computer science educational program. An extended technology acceptance model (TAM) that includes output quality and perceived impact as external variables has been specified and tested with a sample of 165 university students. The results show that perceived enjoyment and perceived usefulness have the strongest direct influence on the behavioural intention to use ChatGPT. The influence of perceived ease of use on perceived usefulness and behavioural intention was not significant, suggesting that it is not a relevant variable for computer science students. Another finding is that taking into consideration indirect effects, perceived impact of AI tools on education and future professions, and perceived enjoyment emerge as equally important for the adoption of ChatGPT. Keywords: ChatGPT, computer science, university students, TAM, AI tools
- Open Access
- 21 March 2026
- Published by Springer Nature in Acta Universitatis Sapientiae, Informatica
- Vol. 18 (1) , 11
Automatic Detection of Core Sections in Scientific Papers
Academic research workflows can be significantly accelerated by tools that provide targeted access to the most relevant parts of scientific articles, avoiding the need to read full documents. Since interest typically centres on key elements such as the research problem, proposed approach, and main findings, automatic detection of a paper’s structural sections can greatly facilitate navigation and content retrieval. In this work, we address the task of sentence-level section classification in research papers, assigning each sentence to categories such as Introduction , Related Work , Proposed Approach, Body, Results, and Conclusion . We propose a classification pipeline built on a high-quality dataset constructed via a dedicated filtering and selection process. Two pre-trained models are first used to automatically label sentences extracted from research articles; only sentences for which both models agree on the predicted label are retained, ensuring consistent annotation. On this refined dataset, we fine-tune a BERT-Base* classifier that achieves an average precision of 84.67%, recall of 83.78%, F1 score of 84.86%, and an overall accuracy of 86.49%, indicating robust and well-balanced performance across classes. The dataset, model, and implementation code are publicly available. By enabling accurate sentence-level structural tagging, our approach supports more efficient research workflows and lays the groundwork for advanced applications in summarisation, information retrieval, and semantic search over large collections of scientific documents.
- 21 March 2026
- Published by Springer Nature in Acta Universitatis Sapientiae, Informatica
- Vol. 18 (1) , 11
Open AccessAutomatic Detection of Core Sections in Scientific Papers
Academic research workflows can be significantly accelerated by tools that provide targeted access to the most relevant parts of scientific articles, avoiding the need to read full documents. Since interest typically centres on key elements such as the research problem, proposed approach, and main findings, automatic detection of a paper’s structural sections can greatly facilitate navigation and content retrieval. In this work, we address the task of sentence-level section classification in research papers, assigning each sentence to categories such as Introduction , Related Work , Proposed Approach, Body, Results, and Conclusion . We propose a classification pipeline built on a high-quality dataset constructed via a dedicated filtering and selection process. Two pre-trained models are first used to automatically label sentences extracted from research articles; only sentences for which both models agree on the predicted label are retained, ensuring consistent annotation. On this refined dataset, we fine-tune a BERT-Base* classifier that achieves an average precision of 84.67%, recall of 83.78%, F1 score of 84.86%, and an overall accuracy of 86.49%, indicating robust and well-balanced performance across classes. The dataset, model, and implementation code are publicly available. By enabling accurate sentence-level structural tagging, our approach supports more efficient research workflows and lays the groundwork for advanced applications in summarisation, information retrieval, and semantic search over large collections of scientific documents.
- 22 June 2025
- Published by Springer Nature
- p. 383-396
Optimizing Cache Utilization in CPU Compilation Through Compiler Autotuning
- 22 June 2025
- Published by Springer Nature
- p. 383-396
Optimizing Cache Utilization in CPU Compilation Through Compiler Autotuning
- 22 June 2025
- Published by Springer Nature
- p. 421-433
OPJUSTICE - A Multi-label Text Classification Model
- 22 June 2025
- Published by Springer Nature
- p. 421-433
OPJUSTICE - A Multi-label Text Classification Model
- 14 November 2024
- Published by Springer Nature
- p. 25-34
Hotel’s Price Prediction Based on Country Specific Data
- 14 November 2024
- Published by Springer Nature
- p. 25-34
Hotel’s Price Prediction Based on Country Specific Data
- 14 November 2024
- Published by Springer Nature
- p. 15-24
Smart Sign Language Decoder
- 10 April 2024
- Published by Springer Nature
- p. 187-203
Evaluation of the Driving Behaviour Models with Newly Collected Data
- 10 April 2024
- Published by Springer Nature
- p. 187-203
Evaluation of the Driving Behaviour Models with Newly Collected Data
- Open Access
- 21 April 2022
- Published by MDPI AG in Electronics
- Vol. 11 (9) , 1316
Building and Using Multiple Stacks of Models for the Classification of Learners and Custom Recommending of Quizzes
Recommending quizzes in e-Learning systems always represents a challenging task, as the quality of recommendations may have a high impact on the student’s progress. We propose a data analysis workflow based on building multiple stacks of models that use information from former students’ taken quizzes. The current implementation uses the RandomForest algorithm for building the models on a real-world dataset that has been obtained in a controlled environment. As preprocessing techniques, we have used normalization and discretization such that training data have been used for classification and regression tasks. At run-time, the models are queried for classifying the student and inferring an optimal quiz that is personalized for the student. We have evaluated the accuracy parametrized on the previous number of quizzes and found that a possible optimal timeframe for each class of students should be used and may provide more helpful quizzes.
- 21 April 2022
- Published by MDPI AG in Electronics
- Vol. 11 (9) , 1316
Open AccessBuilding and Using Multiple Stacks of Models for the Classification of Learners and Custom Recommending of Quizzes
Recommending quizzes in e-Learning systems always represents a challenging task, as the quality of recommendations may have a high impact on the student’s progress. We propose a data analysis workflow based on building multiple stacks of models that use information from former students’ taken quizzes. The current implementation uses the RandomForest algorithm for building the models on a real-world dataset that has been obtained in a controlled environment. As preprocessing techniques, we have used normalization and discretization such that training data have been used for classification and regression tasks. At run-time, the models are queried for classifying the student and inferring an optimal quiz that is personalized for the student. We have evaluated the accuracy parametrized on the previous number of quizzes and found that a possible optimal timeframe for each class of students should be used and may provide more helpful quizzes.
- 23 March 2022
- Published by Springer Nature
- p. 81-99
Enhancing Machine Learning Models by Augmenting New Functionalities
- 23 March 2022
- Published by Springer Nature
- p. 81-99
Enhancing Machine Learning Models by Augmenting New Functionalities