MVMihai Vlase

Mihai Vlase

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Mihai Vlase has a total of 18 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    RubricAdapt-LLM: Measuring Criterion-Level Adaptation and Score Shifts Under Alternative Grading Rubrics

    by Catalin Anghel, Andreea Alexandra Anghel, Adina Cocu, Marian Viorel Craciun, Simona Moldovanu, Mihai Vlase, Cristian Sandu

    Background: Large language models are increasingly used for rubric-based grading, but it remains unclear whether they adapt selectively when criterion weights change while the response, criterion definitions, and total score remain fixed. This study examined whether alternative point allocations produce targeted criterion-level adaptation or broader grading instability. Methods: A controlled paired design was applied to 1000 student responses, comprising 500 technical and 500 argumentative answers. Eight local open-weight LLMs evaluated every response under two analytic rubrics totaling 10 points. One point was transferred from Clarity to Completeness for technical responses and from Clarity to Dialecticality for argumentative responses. Of 16,000 expected evaluations, 15,999 were structurally valid, yielding 7999 complete cross-rubric pairs. Model outputs were analyzed for total-score shifts, affected-criterion adaptation, stability of unaffected criteria, model–human alignment, and correspondence with differences between two human evaluation conditions. Results: All models assigned lower mean scores under Rubric B, with mean shifts ranging from −1.239 to −0.182 points. Adaptation mechanisms differed substantially across models and response types. gemma3:4b frequently preserved technical total scores through compensating criterion changes, whereas llama3.1:8b showed extensive spillover into unaffected criteria. The Qwen models generally produced smaller total-score reductions and greater stability in unchanged dimensions. The mean score was 0.637 points higher under the human Rubric B condition than under the human Rubric A condition, although the two conditions were applied by different evaluator pairs; every model shifted negatively, and the lowest overall shift error was obtained by qwen3:4b at 1.498 points. Conclusions: Rubric sensitivity did not consistently imply localized or criterion-consistent adaptation. Reliable evaluation of LLM graders therefore requires separate analysis of total scores, affected criteria, unaffected criteria, human alignment, and cross-condition shift correspondence.


  • Open Access

    ConsensusGrade: A Human-Variability-Aware Framework for Evaluating LLM-Based Automated Grading

    by Catalin Anghel, Andreea Alexandra Anghel, Mihai Vlase, Marian Viorel Craciun, Adina Cocu, Constantin Adrian Andrei, Diana-Elena Vulpe, Serban Dragosloveanu, Cristian Scheau, Calina Maier, et al.

    Background: Evaluation of LLM-based automated grading often relies on comparison with a single human score, which can obscure meaningful variability among raters of open-ended answers. This study introduces ConsensusGrade, a consensus-aware framework that treats the human reference as a scoring envelope rather than as a single point. Methods: We analyzed 1000 open-ended student answers from 100 students across 10 questions, each graded by four evaluators. Six previously generated and aligned automated grading configurations from GradeAgentOps were compared with the four-rater human reference. The score sets were generated using Llama 3.3 70B Instruct as the primary grader, with Qwen 2.5 14B Instruct for semantic repair. Results: Human evaluators showed meaningful agreement, with ICC(A,1) = 0.712, but exact four-rater agreement occurred in only 2.2% of records. Broad score dispersion occurred in 59.0%. All automated configurations showed negative bias relative to the human median. FULL achieved 68.5% inside-envelope positioning and a chance-adjusted score of 0.454; under the central-trimmed envelope, this rate decreased to 34.3%, while configuration ordering was preserved. Conclusions: ConsensusGrade provides a diagnostic framework for interpreting automated scores relative to observed human variability; inside-envelope rates should not be interpreted as stand-alone measures of grading accuracy.

  • Open Access

    Navigating Learning Paths with EDSense: An AI-Powered Learning Platform

    by Marius Ivanov, Dragoș Sebastian Cristea, Mihai Vlase, Dan Munteanu

    In the realm of contemporary education, the integration of advanced technology is not just a trend but a necessity. ”EDSense” emerges as a cutting-edge platform at the intersection of educational needs and technological innovation. This paper focuses on the development of a social educational platform, leveraging artificial intelligence (AI) uniquely designed to support the official educational curriculum, catering to a wide range of users from students and teachers in grades I-XII in Romania, to adult learners. Central to EDSense is the use of AI for personalized learning experiences. This includes the implementation of sophisticated recommending systems that adapt to individual learning styles and progress. Personalized quizzes, crafted using AI algorithms, offer a tailored approach to assess and reinforce learning. Furthermore, EDSense employs sentiment analysis to gauge and respond to the emotional and cognitive states of learners, thereby enhancing the overall learning experience. The platform stands as a testament to the potential of AI in revolutionizing education. By combining educational content with state-of-the-art technology, EDSense aims to provide an engaging, effective, and highly personalized learning journey, setting a new standard in educational technology.


  • Open Access

    ED-Sense, platformă tehnologică pentru educație personalizată în era digitală bazată pe inteligență artificială și conținut interactiv

    by Marius Petruț Ivanov, Ana-Karina Nazare, Alexandru Stanciu, Dragoș Sebastian Cristea, Mihai Vlase

    The article analyses the current platforms for digital education, identifying the key needs of participants in the educational process and presenting innovative solutions, based on recent research and well-founded theories, to respond to these needs effectively. The EdSense platform addresses the challenges of digitizing education with a creative approach, integrating artificial intelligence and augmented reality to create a ubiquitous, accessible and personalized learning environment. Gamification elements and augmented reality enrich the learning experience, turning it into an engaging and collaborative adventure. Adapting flexibly to the school curriculum and individual student needs through its artificial intelligence component, it offers an adaptive, interactive and deeply personalized learning style. The analysis of similar approaches, as well as the integrative, complex, but well-modularized architecture of the system, profiles EdSense as a reference model for the future of digital education.


  • Open Access

    Digitalised Interactive Educational Experiences with EdSense's Digital Smart Pages

    by Mihai Vlase, Dragoș Sebastian Cristea, Marius Petruț Ivanov

    Recently, there has been a significant evolution in digital educational content that involves transitioning from the simple presentation of information in written form to the inclusion of innovative technology. This alteration is intended to facilitate the ease of integrating educational content. This article describes the concept of Digital Smart Pages (DSP) within the EdSense educational platform context and proposes a design for this educational component. By observing user actions within the DSP and utilizing various components to process these actions, users can have a customized experience within the platform, experience that will lead to a more accessible and enjoyable assimilation of educational content.