Mladen Raković
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Mladen Raković has a total of 140 co-authors. Here are the top co-authors by number of shared publications:
- Dragan Gašević
Centre for Learning Analytics, Faculty of Information Technology, Monash University, Melbourne, Australia
31 shared publications - Guanliang Chen
Monash University
18 shared publications - Yizhou Fan
Peking University
15 shared publications - Xinyu Li
Monash University
13 shared publications - Tongguang Li
Monash University
10 shared publications Dragan Gašević
9 shared publications
Latest Publications
- Open Access
- 21 September 2026
- Published by Wiley in Journal of Computer Assisted Learning
- Vol. 42 (5)
Trace Data of Secondary Students and Linguistic Analysis to Predict Learner Performance in a Multi‐Text Writing Task
Background: Writing from multiple sources is a cognitively demanding task that requires students to integrate reading, planning, and composition processes. While prior research has demonstrated the importance of self‐regulated learning (SRL) and linguistic features in predicting writing performance, most studies have focused on university students, with limited research targeting secondary school learners, particularly across multilingual contexts. Objective: This study examined whether SRL processes derived from trace data and linguistic features from essays as learning products can be used to predict writing performance among secondary students from different countries and language backgrounds. Methods: A total of 257 secondary students from Australia, Colombia, Germany, and Finland completed a multi‐source essay task in their school language within a digital learning environment. SRL processes were inferred from interaction trace data, while essay features—topic coverage, cohesion, and word count—were extracted using language‐specific analysis models. These data sources were then used to train several machine learning models to predict students' performance on essay writing. Results: Among the machine learning models tested, the Random Forest classifier achieved the highest accuracy (0.80) and recall (0.71). The most influential predictors were elaboration/organisation (process), topic coverage and cohesion (product), and country. Results revealed distinct engagement patterns across countries, suggesting that language and educational context were associated with differences in both SRL processes and writing quality. Conclusions: This study extends prior work by demonstrating that combining process and product features can support the prediction of writing performance across multilingual secondary education contexts. Findings emphasise the need for further research on younger learners and the role of linguistic background in SRL engagement.
- 21 September 2026
- Published by Wiley in Journal of Computer Assisted Learning
- Vol. 42 (5)
Open AccessTrace Data of Secondary Students and Linguistic Analysis to Predict Learner Performance in a Multi‐Text Writing Task
Background: Writing from multiple sources is a cognitively demanding task that requires students to integrate reading, planning, and composition processes. While prior research has demonstrated the importance of self‐regulated learning (SRL) and linguistic features in predicting writing performance, most studies have focused on university students, with limited research targeting secondary school learners, particularly across multilingual contexts. Objective: This study examined whether SRL processes derived from trace data and linguistic features from essays as learning products can be used to predict writing performance among secondary students from different countries and language backgrounds. Methods: A total of 257 secondary students from Australia, Colombia, Germany, and Finland completed a multi‐source essay task in their school language within a digital learning environment. SRL processes were inferred from interaction trace data, while essay features—topic coverage, cohesion, and word count—were extracted using language‐specific analysis models. These data sources were then used to train several machine learning models to predict students' performance on essay writing. Results: Among the machine learning models tested, the Random Forest classifier achieved the highest accuracy (0.80) and recall (0.71). The most influential predictors were elaboration/organisation (process), topic coverage and cohesion (product), and country. Results revealed distinct engagement patterns across countries, suggesting that language and educational context were associated with differences in both SRL processes and writing quality. Conclusions: This study extends prior work by demonstrating that combining process and product features can support the prediction of writing performance across multilingual secondary education contexts. Findings emphasise the need for further research on younger learners and the role of linguistic background in SRL engagement.
- Open Access
- 27 July 2026
- Published by Wiley in Journal of Computer Assisted Learning
- Vol. 42 (5) , e70300
Human or GenAI Support? Conditions Impacting Students' Strategy Choices in an Essay Revision Task
Background: Students often struggle to effectively self‐regulate their learning strategies during writing tasks and typically require external support from teachers, peers or AI tools to overcome these challenges. While emerging research explores how generative AI can dialogically interact with students to support the writing process, little is known about how such AI support influences writing strategies during the essay‐revision task. Furthermore, there is limited understanding of how GenAI support compares to other external conditions (human feedback or no support) in shaping writing strategies and how it interacts with internal learning factors such as writing skills, metacognitive accuracy and learning motivation. Objectives: Our study aims to investigate the relationship of self‐regulated learning (SRL) strategies that students adopt for an essay‐revision task with learning conditions (both internal and external) and learning performance. Methods: Using a technology‐enhanced learning environment, we collected trace data from 87 participants during essay‐writing and revision tasks. We applied Mixture Markov Models (MMMs) to model the writing/revision processes, then used the resulting transition matrix within the Expectation–Maximization (EM) clustering algorithm to identify SRL strategies. A series of statistical tests were applied to examine the relationship of strategies that students adopt for a revision task with learning conditions (internal and external) and learning performance. Results and Conclusions: We found moderate associations between writing strategies during the essay‐revision task and internal conditions (writing strategies adopted in the preceding essay‐writing task) and large associations with external conditions (external support). However, no association was found between writing strategies during the essay‐revision task and learning performance. These findings inform the development of educational interventions to improve students' writing strategies.
- 27 July 2026
- Published by Wiley in Journal of Computer Assisted Learning
- Vol. 42 (5) , e70300
Open AccessHuman or GenAI Support? Conditions Impacting Students' Strategy Choices in an Essay Revision Task
Background: Students often struggle to effectively self‐regulate their learning strategies during writing tasks and typically require external support from teachers, peers or AI tools to overcome these challenges. While emerging research explores how generative AI can dialogically interact with students to support the writing process, little is known about how such AI support influences writing strategies during the essay‐revision task. Furthermore, there is limited understanding of how GenAI support compares to other external conditions (human feedback or no support) in shaping writing strategies and how it interacts with internal learning factors such as writing skills, metacognitive accuracy and learning motivation. Objectives: Our study aims to investigate the relationship of self‐regulated learning (SRL) strategies that students adopt for an essay‐revision task with learning conditions (both internal and external) and learning performance. Methods: Using a technology‐enhanced learning environment, we collected trace data from 87 participants during essay‐writing and revision tasks. We applied Mixture Markov Models (MMMs) to model the writing/revision processes, then used the resulting transition matrix within the Expectation–Maximization (EM) clustering algorithm to identify SRL strategies. A series of statistical tests were applied to examine the relationship of strategies that students adopt for a revision task with learning conditions (internal and external) and learning performance. Results and Conclusions: We found moderate associations between writing strategies during the essay‐revision task and internal conditions (writing strategies adopted in the preceding essay‐writing task) and large associations with external conditions (external support). However, no association was found between writing strategies during the essay‐revision task and learning performance. These findings inform the development of educational interventions to improve students' writing strategies.
- 30 June 2026
- Published by Taylor & Francis in College Teaching
- p. 1-16
Teaching Students to Set Goals: A Study Diary Intervention to Support Motivation and Metacognition
- 30 June 2026
- Published by Taylor & Francis in College Teaching
- p. 1-16
Teaching Students to Set Goals: A Study Diary Intervention to Support Motivation and Metacognition
- 27 June 2026
- Published by Springer Nature
- p. 538-544
Capturing Momentary Motivational States Using Detector-Driven In-Situ Questionnaire
- 27 June 2026
- Published by Springer Nature
- p. 538-544
Capturing Momentary Motivational States Using Detector-Driven In-Situ Questionnaire
- 25 June 2026
- Published by Springer Nature
- p. 609-617
Predicting the Finish Before the Draft Ends: Continuous Forecasting of Writing Performance from Process Traces
- 25 June 2026
- Published by Springer Nature
- p. 609-617
Predicting the Finish Before the Draft Ends: Continuous Forecasting of Writing Performance from Process Traces
- Open Access
- 5 June 2026
- Published by Center for Open Science in PsyArXiv
Motivation Science in Digital and AI-Supported Learning Environments
In this conceptual paper, we argue that motivation science provides a valuable yet underutilized framework for understanding and designing digital and AI-supported learning environments. To illustrate this, we first review studies included in a recent meta-analysis on generative AI and motivation to examine the extent to which motivation science informs current research. We show that the vast majority of studies do not use motivational theories to explain how such systems may support student motivation. Next, drawing primarily on achievement goal theory and the TARGET (task, autonomy, grouping, recognition, time) framework, we outline key motivational principles for structuring learning environments and discuss how these principles may be enacted in digital and AI-supported contexts. We also examine risks and tensions associated with these environments, acknowledging that digital and AI systems may create novel motivational challenges not fully addressed by existing motivational theories developed in more traditional educational settings. Overall, the paper aims to stimulate further research on rethinking digital and AI-supported learning environments through the lens of motivation science while also reconsidering motivation science in light of these emerging environments.
- 5 June 2026
- Published by Center for Open Science in PsyArXiv
Open AccessMotivation Science in Digital and AI-Supported Learning Environments
In this conceptual paper, we argue that motivation science provides a valuable yet underutilized framework for understanding and designing digital and AI-supported learning environments. To illustrate this, we first review studies included in a recent meta-analysis on generative AI and motivation to examine the extent to which motivation science informs current research. We show that the vast majority of studies do not use motivational theories to explain how such systems may support student motivation. Next, drawing primarily on achievement goal theory and the TARGET (task, autonomy, grouping, recognition, time) framework, we outline key motivational principles for structuring learning environments and discuss how these principles may be enacted in digital and AI-supported contexts. We also examine risks and tensions associated with these environments, acknowledging that digital and AI systems may create novel motivational challenges not fully addressed by existing motivational theories developed in more traditional educational settings. Overall, the paper aims to stimulate further research on rethinking digital and AI-supported learning environments through the lens of motivation science while also reconsidering motivation science in light of these emerging environments.
- 8 December 2025
- Published by Elsevier in Computers & Education
- Vol. 243
FLoRA: An advanced AI-powered engine to facilitate hybrid human-AI regulated learning
- 8 December 2025
- Published by Elsevier in Computers & Education
- Vol. 243
FLoRA: An advanced AI-powered engine to facilitate hybrid human-AI regulated learning
- 26 November 2025
- Published by Wiley in British Journal of Educational Technology
- Vol. 57 (2) , 508-534
Ink and algorithm: Exploring temporal dynamics in generative AI ‐assisted writing
The advent of Generative AI (GAI) has transformed writing, marking a shift towards GAI‐assisted writing in education. However, the dynamics of human–AI interaction in the writing process are not well‐understood, and thus, it remains largely unknown how human learning can be effectively supported with such technologies. This study addresses this gap by investigating how humans employ GAI during writing and examining the interplay between patterns of GAI usage and writing behaviours. To capture these patterns, we applied Dynamic Time Warping time‐series clustering to identify temporal trajectories of GAI use and employed Epistemic Network Analysis to examine how these trajectories relate to cognitive processes such as knowledge telling, knowledge transformation and cognitive presence. Our analysis revealed four distinct temporal patterns of GAI usage (ie, AI‐critical writers, AI‐dependent writers, AI‐independent writers and AI‐balanced writers), each associated with different cognitive engagement strategies. The findings suggest that some writers tend to rely excessively on GAI, which may limit opportunities for meaningful learning by reinforcing surface‐level strategies such as knowledge telling. These insights highlight the need for researchers and educators to design GAI‐based writing assistants that are sensitive to students' temporal dynamics and can scaffold more productive cognitive engagement. Practitioner notes: What is already known about this topic Researchers and educators can gain practical insights into achieving intelligence augmentation through critical engagement by studying effective user behaviours for enhanced human–AI partnership in writing. Generative AI‐assisted writing can be evaluated using an evidence‐centred assessment framework, which relates to writing cognitive processes. Current studies on Generative AI‐assisted writing usually classify writers according to their overall AI usage behaviours throughout the entire writing session. What this paper adds We propose using time‐series clustering to identify and analyse common temporal patterns in AI usage during generative AI‐assisted writing processes. We uncover the correlation between temporal patterns in AI usage and human writing behaviours, which reflect cognitive processes, through Epistemic Network Analysis. We identify four major distinct temporal patterns in AI utilization and highlight that each pattern is correlated with different cognitive processes. Implications for practice and/or policy Researchers and educators should be aware of the risks associated with students overly relying on GAI for writing tasks, as this dependency can hinder opportunities for meaningful learning. Researchers and educators should carefully consider the types of suggestions provided by GAI‐based writing tools when integrating them into educational contexts.
- 26 November 2025
- Published by Wiley in British Journal of Educational Technology
- Vol. 57 (2) , 508-534
Ink and algorithm: Exploring temporal dynamics in generative AI ‐assisted writing
The advent of Generative AI (GAI) has transformed writing, marking a shift towards GAI‐assisted writing in education. However, the dynamics of human–AI interaction in the writing process are not well‐understood, and thus, it remains largely unknown how human learning can be effectively supported with such technologies. This study addresses this gap by investigating how humans employ GAI during writing and examining the interplay between patterns of GAI usage and writing behaviours. To capture these patterns, we applied Dynamic Time Warping time‐series clustering to identify temporal trajectories of GAI use and employed Epistemic Network Analysis to examine how these trajectories relate to cognitive processes such as knowledge telling, knowledge transformation and cognitive presence. Our analysis revealed four distinct temporal patterns of GAI usage (ie, AI‐critical writers, AI‐dependent writers, AI‐independent writers and AI‐balanced writers), each associated with different cognitive engagement strategies. The findings suggest that some writers tend to rely excessively on GAI, which may limit opportunities for meaningful learning by reinforcing surface‐level strategies such as knowledge telling. These insights highlight the need for researchers and educators to design GAI‐based writing assistants that are sensitive to students' temporal dynamics and can scaffold more productive cognitive engagement. Practitioner notes: What is already known about this topic Researchers and educators can gain practical insights into achieving intelligence augmentation through critical engagement by studying effective user behaviours for enhanced human–AI partnership in writing. Generative AI‐assisted writing can be evaluated using an evidence‐centred assessment framework, which relates to writing cognitive processes. Current studies on Generative AI‐assisted writing usually classify writers according to their overall AI usage behaviours throughout the entire writing session. What this paper adds We propose using time‐series clustering to identify and analyse common temporal patterns in AI usage during generative AI‐assisted writing processes. We uncover the correlation between temporal patterns in AI usage and human writing behaviours, which reflect cognitive processes, through Epistemic Network Analysis. We identify four major distinct temporal patterns in AI utilization and highlight that each pattern is correlated with different cognitive processes. Implications for practice and/or policy Researchers and educators should be aware of the risks associated with students overly relying on GAI for writing tasks, as this dependency can hinder opportunities for meaningful learning. Researchers and educators should carefully consider the types of suggestions provided by GAI‐based writing tools when integrating them into educational contexts.
- Open Access
- 18 November 2025
- Published by Center for Open Science in PsyArXiv
The Impact of Scaffold Interactions on Learning: A Learning Analytics Approach using Multimodal Data
Scaffolds are widely used to support self-regulated learning (SRL), and research has shown that students’ interactions with scaffolds impact their effectiveness. Prior research has primarily classified scaffold interaction as a binary construct (e.g., compliance vs non-compliance), which oversimplifies complex scaffold interaction patterns and limits understanding of the effectiveness of scaffolds. The present study addressed this limitation by adopting a more nuanced learning analytic perspective to examine the range and quality of scaffold interaction using multimodal data. In a pre-post-test experiment, university students (n = 67) learned with adaptive scaffolds—differentiated as just-in-time or on-demand—in a 45-minute learning session using a technology-enhanced learning environment. Combining logfile and eye-tracking data, we identified distinct interaction profiles through hierarchical agglomerative clustering of students’ scaffold interactions over the course of learning. Learners who engaged more consistently and actively with scaffolds demonstrated higher levels of cognitive activity changes and achieved stronger performance in essay tasks than peers with minimal or purely visual interaction (i.e., sighting but not acting on scaffolds). The process model comparison highlighted differences in the temporal structure of learning and scaffold interaction activities between groups with varying interaction profiles. These findings highlight the shortcomings of binary classifications and provide new insights into how adaptive scaffolds can be designed to respond to diverse interaction patterns and better foster SRL.
- 18 November 2025
- Published by Center for Open Science in PsyArXiv
Open AccessThe Impact of Scaffold Interactions on Learning: A Learning Analytics Approach using Multimodal Data
Scaffolds are widely used to support self-regulated learning (SRL), and research has shown that students’ interactions with scaffolds impact their effectiveness. Prior research has primarily classified scaffold interaction as a binary construct (e.g., compliance vs non-compliance), which oversimplifies complex scaffold interaction patterns and limits understanding of the effectiveness of scaffolds. The present study addressed this limitation by adopting a more nuanced learning analytic perspective to examine the range and quality of scaffold interaction using multimodal data. In a pre-post-test experiment, university students (n = 67) learned with adaptive scaffolds—differentiated as just-in-time or on-demand—in a 45-minute learning session using a technology-enhanced learning environment. Combining logfile and eye-tracking data, we identified distinct interaction profiles through hierarchical agglomerative clustering of students’ scaffold interactions over the course of learning. Learners who engaged more consistently and actively with scaffolds demonstrated higher levels of cognitive activity changes and achieved stronger performance in essay tasks than peers with minimal or purely visual interaction (i.e., sighting but not acting on scaffolds). The process model comparison highlighted differences in the temporal structure of learning and scaffold interaction activities between groups with varying interaction profiles. These findings highlight the shortcomings of binary classifications and provide new insights into how adaptive scaffolds can be designed to respond to diverse interaction patterns and better foster SRL.