HKHassan Khosravi

Hassan Khosravi

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

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


Hassan Khosravi has a total of 207 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Building AI companions that prioritise learning over performance

    by Hassan Khosravi, Dragan Gašević, Shazia Sadiq, Lixiang Yan, Jason M. Lodge, Jason M. Tangen, Paul Denny, Kristen DiCerbo, Simon Buckingham Shum, Ryan S. Baker


  • Open Access

    Retrieval-Augmented Contrastive Learning for Knowledge Tracing

    by Kamal Berahmand, Mehrnoush Mohammadi, Homa Babai, Hassan Khosravi


  • Top Cited Papers
    Open Access

    AI assistance in peer feedback provision: Pedagogically sound, but minimally adopted

    by Stanislav Pozdniakov, Jonathan Brazil, Seyyed Kazem Banihashem, Omid Noroozi, Dragan Gašević, Shazia Sadiq, Hassan Khosravi


  • Top Cited Papers
    Open Access

    Generative AI offers more, but students revise less: comparing the effects of teacher and AI feedback on student essay revisions

    by Mohammadreza Farrokhnia, Saeed Latifi, Pantelis M. Papadopoulos, Loes Hogenkamp, Hannie Gijlers, Hassan Khosravi, Omid Noroozi

    Providing high-quality feedback on student writing is essential yet increasingly difficult due to rising class sizes and limited instructional capacity. Generative AI (GenAI) offers a promising and scalable alternative, but its effectiveness compared to traditional teacher feedback, particularly across different prompting techniques, remains uncertain. This study employed a quantitative, randomized three-group experimental design with 70 graduate students to compare the effects of teacher feedback and GenAI feedback generated using two prompting techniques: Zero-shot and chain-of-thought (CoT). The study explored how these feedback sources affect feedback quality and students’ uptake during essay revision. It involved a two-stage process in which students first wrote an argumentative essay and then revised it based on the feedback received. Feedback and essay quality were evaluated using a rubric based on Toulmin’s model of argumentation and analysed using inferential statistical methods. Results showed that CoT prompting produced higher quality feedback than both Zero-shot prompting and teacher feedback, suggesting that stepwise reasoning in CoT aligns GenAI outputs more closely with the cognitive demands of argumentative writing. However, this higher feedback quality did not lead to significantly greater improvements in revisions. Teacher feedback, although rated lower in quality, resulted in comparable gains in essay quality. In addition, GenAI feedback quality was significantly associated with students’ initial essay quality, whereas teacher feedback quality showed no such association. These findings indicate that feedback quality alone is insufficient to enhance writing outcomes; rather, students’ engagement with and uptake of feedback play a critical role. Overall, the results highlight the potential of hybrid intelligent feedback systems in which teachers support students in interpreting and applying GenAI feedback to meaningfully improve their writing.


  • Open Access

    More than meets the mind's eye: vividness of spatial mental imagery as a predictor of stereochemistry performance

    by Lauren Baade, Effie Kartsonaki, Hassan Khosravi, Gwendolyn Lawrie

    During early organic chemistry learning, students often rely on imagistic reasoning to solve visuospatial problems. In chemistry education research, such problem solving is commonly assumed to rely on visual mental imagery. Individual differences in visual mental imagery vividness – including aphantasia, the absence of visual mental imagery held by an individual – motivate a broader question: whether visual mental imagery alone captures the imagery-related capacities relevant to spatially demanding problem solving. This study examined how undergraduate students’ self-reported visual mental imagery vividness could be distinguished from a broader spatially oriented dimension of mental imagery, and whether these dimensions were differently associated with introductory stereochemistry performance. 172 first-year chemistry students completed a six-item Foundational Stereochemistry Skills Instrument (FoSSI) alongside a Spatial and Visual Mental Imagery Questionnaire (SAVMIQ) developed for this purpose. The SAVMIQ showed preliminary evidence of discriminating spatial and visual mental imagery vividness, with factor-analytic results supporting a two-factor structure and the subscales showing differing relationships with stereochemistry accuracy. Higher SAVMIQ spatial mental imagery vividness scores were associated with greater accuracy across five of the six FoSSI items, whereas SAVMIQ visual mental imagery vividness scores were negatively associated with accuracy on one enantiomer-identification item. The findings suggest that success in introductory stereochemistry is more closely associated with students’ reported vividness for imagined spatial operations than with their vividness of visual mental imagery. These findings contribute to understandings of imagistic reasoning by suggesting that visual vividness alone may not capture the imagery-related capacities associated with stereochemistry problem solving.

  • Open Access

    Fostering Human Agency in Age of AI

    by Olga Viberg, Oleksandra Poquet, Vitomir Kovanovic, Hassan Khosravi

    As learning analytics (LA) and artificial intelligence (AI) increasingly shape how learning processes are monitored and supported, human agency has emerged as a critical concern. With generative AI rapidly transforming learning practices and influencing pedagogical decision-making, safeguarding the agency of both learners and educators is becoming essential. This editorial discusses key dilemmas in designing and evaluating AI-mediated learning systems that maintain meaningful human control, foster critical engagement, and enable ethical and effective integration into learning settings. We conclude by outlining future research opportunities for the learning analytics community and reflecting on the journal’s development over the past year.


  • Open Access

    From knowledge gaps to learning opportunities: Leveraging student questions and dual use of generative AI to support student learning at scale

    by Stanislav Pozdniakov, Jonathan Brazil, Oleksandra Poquet, Stephan Krusche, Santiago Berrezueta-Guzman, Shazia Sadiq, Hassan Khosravi


  • Open Access

    Directive, Metacognitive or a Blend of Both? A Comparison of AI-Generated Feedback Types on Student Engagement, Confidence, and Outcomes

    by Omar Alsaiari, Nilufar Baghaei, Jason M. Lodge, Omid Noroozi, Dragan Gašević, Marie Boden, Hassan Khosravi

    Feedback is one of the most powerful influences on student learning, with extensive research examining how best to implement it in educational settings. Increasingly, feedback is being generated by artificial intelligence (AI), offering scalable and adaptive responses. Two widely studied approaches are directive feedback, which gives explicit explanations and reduces cognitive load to speed up learning, and metacognitive feedback which prompts learners to reflect, track their progress, and develop self-regulated learning (SRL) skills. While both approaches have clear theoretical advantages, their comparative effects on engagement, confidence, and quality of work remain underexplored. This study presents a semester-long randomised controlled trial with 329 students in an introductory design and programming course using an adaptive educational platform. Participants were assigned to receive directive, metacognitive, or hybrid AI-generated feedback that blended elements of both directive and metacognitive feedback. Results showed that revision behaviour differed across feedback conditions, with Hybrid prompting the most revisions compared to Directive and Metacognitive. Confidence ratings were uniformly high, and resource quality outcomes were comparable across conditions. These findings highlight the promise of AI in delivering feedback that balances clarity with reflection. Hybrid approaches, in particular, show potential to combine actionable guidance for immediate improvement with opportunities for self-reflection and metacognitive growth.


  • Open Access

    Relative Entropy-based Regularized Non-negative Matrix Factorization for Attributed Graph Clustering

    by Kamal Berahmand, Mehrnoush Mohammadi, Razieh Sheikhpour, Mahdi Jalili, Richi Nayak, Hassan Khosravi

    Attributed graph clustering is a fundamental task in network mining, essential for uncovering valuable insights in various applications. However, the heterogeneity of information from structural and attribute spaces poses significant challenges in achieving consistent and meaningful clustering. To address this, we propose Relative Entropy-based Regularized Non-negative Matrix Factorization (RENMF), a novel approach that integrates structural and attribute information through advanced matrix factorization techniques. RENMF employs Symmetric NMF and Projective NMF to extract community membership distributions from the structural and attribute spaces, respectively. By treating these distributions as homogeneous, RENMF preserves distinct, denoised information from both spaces while considering their heterogeneous complementary information. We introduce Relative Entropy (RE) as a novel regularization term to facilitate interaction between these spaces, aiming to maximize consistency between the discovered latent distributions. In this interaction, we leverage the asymmetric property of RE to emphasize attributes as essential complementary information for structural clustering. The RENMF model is solved using a new iterative multiplicative update rule, with convergence theoretically proven. We evaluate RENMF’s effectiveness through extensive experiments on 10 real-world networks, comparing it to 11 state-of-the-art clustering methods. The results demonstrate RENMF’s superiority in ground truth matching and key quality metrics, outperforming existing methods.