Towards AI-Powered Personalization in MOOC Learning
Open accessAbstract
Massive Open Online Courses (MOOCs) represent a form of large-scale learning that is changing the landscape of higher education. In this paper, we offer a perspective on how advances in artificial intelligence (AI) may enhance learning and research on MOOCs. We focus on emerging AI techniques including how knowledge representation tools can enable students to adjust the sequence of learning to fit their own needs; how optimization techniques can efficiently match community teaching assistants to MOOC mediation tasks to offer personal attention to learners; and how virtual learning companions with human traits such as curiosity and emotions can enhance learning experience on a large scale. These new capabilities will also bring opportunities for educational researchers to analyse students’ learning skills and uncover points along learning paths where students with different backgrounds may require different help. Ethical considerations related to the application of AI in MOOC education research are also discussed.
AI snapshot
Written from the paper’s full text.
TL;DR
This perspective article argues that AI can make MOOC learning more personalized through knowledge-representation tools for flexible learning paths, optimization for community teaching assistants, and virtual learning companions. It also outlines ethical and technical barriers, including hard-to-define learning outcomes, biased completion data, and unresolved conversational AI and consent issues.
Key findings
- Proposes using micro-MOOCs (short videos typically no longer than 5 min each) plus knowledge-representation tools such as Goal Net/MADE or BDI models to let learners reorder topics when dependencies are weak, while tracking personalized paths in machine-understandable form.
- Proposes stochastic optimization and data-driven algorithmic management to scale community teaching assistants; notes current recruitment is small (typically less than 10 community TAs per course) and that about four out of five community TAs declined a second invitation.
- Proposes virtual learning companions (VLCs) with human-like traits such as curiosity and emotions—citing examples like Jill Watson and Betty’s Brain—to provide large-scale interaction and generate learning-behaviour data, but concludes VLCs are not yet ready.
- Discusses ethical considerations for AI in MOOC research, including Belmont Report principles, informed consent, moral decision-making frameworks, explainable AI, and risks of learners gaming VLC limitations.
Methods
- Perspective/conceptual paper; no empirical design, dataset, sample, or main statistical analysis is reported.
- The authors synthesize prior literature and examples—Goal Net, micro-MOOCs, community TAs, VLCs—to outline AI opportunities for MOOC personalization and research.
Limitations
- Stated: learning outcomes are often hard to define and measure, and without a proper learning-outcome objective function, machine learning cannot compute optimized personal learning paths.
- Stated: available learning-behaviour data mostly come from learners who completed MOOCs, so AI models may be biased toward high-performing learners; the extent of this bias requires more study.
- Stated: VLCs are not ready because meaningful human-like conversation remains an open research challenge and ethical issues require careful deliberation; evident from the text: the proposed framework is not empirically validated in this paper.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 38 cited sources checked · 0 supported · 0 flagged
- Interpretationreasoning
Many MOOCs belong to the category of advanced topics with unclear dependencies.
“As MOOCs are often university level courses, many belong to the latter category.”
Quote in the paperSubstantial claim - Premisecitation
Backward design can be used to develop courses with finely grained coherent learning contents.
“The technique of backward design,12 in which course instructors decide objectives, determine evaluation criteria and then choose learning activities, can be used to develop courses with finely grained coherent learning contents.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [12] Childre, A., Sands, J. R. & Pope, S. T. Backward design: targeting depth of understanding for all learners. Teach. Except. Child. 41, 6–14 (2009). source
Not checked in the free run.
- Premisecitation
Effective learning through MOOCs requires pedagogies different from those used in face-to-face learning.
“Today, it is well-recognized that effective learning through MOOCs requires pedagogies different from those used in face-to-face learning.5”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [5] Knox, J. MOOC pedagogy: the challenges of developing for Coursera. Association for Learning Technology Online Newsletter. http://newsletter.alt.ac.uk/2012/08/mooc–pedagogy–the–challenges–of–developing–for–coursera/ (2012).
Not checked in the free run.
- Premisecitation
MOOC platforms have not yet made learning behaviour data across multiple courses accessible to AI researchers.
“As MOOC platforms (e.g. Coursera, edX) have not yet made learning behaviour data across multiple courses accessible to AI researchers,9”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [9] Reich, J. Rebooting MOOC research. Science 347, 34–35 (2015). source
Not checked in the free run.
- Premisecitation
There is evidence suggesting that educators can benefit from learners' inputs on the sequence of learning contents in e-learning environments.
“There is also evidence suggesting that educators can benefit from learners’ inputs on the sequence of learning contents in e-learning environments.11”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [11] McLoughlin, C. & Lee, M. J. W. The three P’s of pedagogy for the networked society: personalization, participation, and productivity. Int. J. Teach. Learn. High. Educ. 20, 10–27 (2008).
Not checked in the free run.
- Premisecitation
Existing AI research in this area mostly relies on course-specific datasets released in an ad-hoc manner.
“existing AI research in this area mostly relies on course-specific datasets released in an ad-hoc manner (e.g. the KDD Cup 2015 competition dataset).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [9] Reich, J. Rebooting MOOC research. Science 347, 34–35 (2015). source
Not checked in the free run.
- Premisecitation
Techniques such as community TAs, peer assessment, rubrics, video feedback are available to help MOOC instructors scale up provision of perceived personal attention.
“Techniques such as involvement of community teaching assistants (TAs),21 peer assessment tools,22 well-designed elaborated rubrics with specific content feedback23 and video feedbacks24 are available to help MOOC instructors scale up the provision of perceived personal attention to students.”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites [21] Papadopoulos, K., Sritanyaratana, L. & Klemmer, S. R. Community TAs scale high-touch learning, provide student-staff brokering, and build esprit de corps. In L@S, 163–164 (2014). source
Not checked in the free run.
- cites [22] Kulkarni, C. et al. Peer and self assessment in massive online classes. ACM Trans. Comput. Hum. Interact. 20, 1–33 (2013). source
Not checked in the free run.
- cites [23] Stevens, D. D. & Levi, A. J. Introduction to Rubrics: An Assessment Tool to Save Grading Time, Convey Effective Feedback, and Promote Student Learning (Stylus Publishing, LLC, Sterling, Virginia, 2013).
Not checked in the free run.
- Premisecitation
Data from these interactions have not yet been systematically collected and managed to support assessment of learning skills.
“Data from these interactions have not yet been systematically collected and managed to support assessment of learning skills,49 however.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [49] Fidalgo-Blanco, Á., Sein-Echaluce, M. L. & García-Peñalvo, F. J. Methodological approach and technological framework to break the current limitations of MOOC model. J. Univers. Comput. Sci. 21, 712–734 (2015).
Not checked in the free run.
- Premisecitation
It is possible to apply topic-modelling-based machine learning approaches to infer learners' attitude, ability and latent skill levels based on forum discussion data if the forum is actively used.
“It is possible to apply emerging topic-modelling-based machine learning approaches to infer learners’ attitude, ability and latent skill levels based on forum discussion data if the forum is actively used by learners.28”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [28] He, J. et al. MOOCs meet measurement theory: a topic-modelling approach. In AAAI, 1195–1201 (2016). source
Not checked in the free run.
- Interpretationcitation
Interactions between course instructors and students as well as among students could provide opportunities to generate complex behaviour data which can be used to analyse students' learning skills.
“Interactions between course instructors and students as well as among students of the same course could provide opportunities to generate complex behaviour data which can be used to analyse students’ learning skills (e.g. twenty-first century learning skills such as collaboration with others, time management and critical thinking19).”
Quote in the paperSubstantial claim - Interpretationreasoning
VLCs are not yet ready for two main reasons.
“Nevertheless, we believe it is not yet ready for two main reasons.”
Quote in the paperNot a substantial claim - Premisecitation
Machine learning-based learner assessment techniques mainly reflect how students interact with, rather than learn from, the courses.
“Using data derived from interaction mechanisms currently available in MOOC platforms, machine learning-based learner assessment techniques mainly reflect how students interact with, rather than learn from, the courses.9,26,27”
Quote in the paperSubstantial claim0 of 3 sources checked- cites [9] Reich, J. Rebooting MOOC research. Science 347, 34–35 (2015). source
Not checked in the free run.
- cites [26] Champaign, J. et al. Correlating skill and improvement in 2 MOOCs with a student’s time on tasks. In L@S, 11–20 (2014). source
Not checked in the free run.
- cites [27] Wilkowski, J., Deutsch, A. & Russell, D. M. Student skill and goal achievement in the mapping with Google MOOC. In L@S, 3–10 (2014). source
Not checked in the free run.
- Premisecitation
Small scale and poor retention of community TAs limit interactions and attentions provided to MOOC learners.
“These studies found that the small scale and poor retention of community TAs limit interactions and attentions which can be provided to MOOC learners.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [29] Pepicello, W. Game Changers: Education and Information Technologies (Educause, 2012).
Not checked in the free run.
- Interpretationreasoning
Creating more interaction opportunities for learners to exhibit various learning skills would help generate the data needed for AI to accurately assess learners' skills.
“Thus, creating more interaction opportunities for learners to exhibit various learning skills would help generate the data needed for AI to accurately assess learners’ skills.”
Quote in the paperSubstantial claim - Premisecitation
Most community TAs (about four out of five) declined invitations to perform duties for a second time.
“Even so, it has been reported that most community TAs (about four out of five) declined invitations to perform community TA duties for a second time.29”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [29] Pepicello, W. Game Changers: Education and Information Technologies (Educause, 2012).
Not checked in the free run.
- Interpretationreasoning
Without proper definition of a learning outcome objective function, machine learning approaches will not be able to compute optimized personal learning paths.
“Without proper definition of a learning outcome objective function, machine learning approaches will not be able to compute optimized personal learning paths.”
Quote in the paperSubstantial claim - Interpretationreasoning
Currently, data concerning learner interaction behaviours in MOOC tend to be sparse.
“Currently, data concerning learner interaction behaviours in MOOC tend to be sparse.”
Quote in the paperSubstantial claim - Interpretationreasoning
For courses which emphasize skills, the current mechanism of content delivery and assessment in MOOC platforms is not very suitable.
“For such courses in which students are to acquire skills, the current mechanism of content delivery and assessment in MOOC platforms is not very suitable.”
Quote in the paperSubstantial claim - Interpretationreasoning
If GANs were trained on sparse available learning behaviour data, there is still risk that generated data will not capture reality well.
“if they were trained on sparse available learning behaviour data, there is still the risk that the resulting generated data will not capture reality well.”
Quote in the paperNot a substantial claim - Premisecitation
Generative Adversarial Networks can partially address data sparsity by synthesizing artificial training data based on statistical distribution of small real data.
“Although AI techniques such as Generative Adversarial Networks25 can partially address the data sparsity issue in pattern recognition tasks by synthesizing artificial training data based on the statistical distribution of a small amount of real training data,”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [25] Goodfellow, I. et al. Generative adversarial nets. In Advances in Neural Information Processing Systems Vol. 27 (eds. Ghahramani, Z. et al.) 2672–2680 (2014).
Not checked in the free run.
- Classified as probably not a substantial claim. 22
- May combine two claims. 8
Probably not a substantial claim: may not pass S4 (based).
Interpretationnone givenFor courses on some advanced topics, the dependence among topics is not always clearly defined.
“However, for courses on some advanced topics (e.g. human–computer interaction), the dependence among topics is not always clearly defined.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationMicro-MOOCs employ micro-lectures typically no longer than 5 min each, along with concise textual explanations.
“They employ a new form of micro-lectures which typically consist of short videos on specific topics, typically no longer than 5 min each, along with concise textual explanations.5”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [5] Knox, J. MOOC pedagogy: the challenges of developing for Coursera. Association for Learning Technology Online Newsletter. http://newsletter.alt.ac.uk/2012/08/mooc–pedagogy–the–challenges–of–developing–for–coursera/ (2012).
Not checked in the free run.
Classified as probably not a substantial claim.
PremisecitationGoal Net allows users to express a complex process as a network of goals and activities.
“Based on the goal setting theory,15 Goal Net allows users to express a complex process (e.g. organizing a set of learning tasks) as a network of goals and activities.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [13] Yu, H., Shen, Z. & Miao, C. A goal oriented development tool to automate the incorporation of intelligent agents into interactive digital media applications. ACM Comput. Entertain. 6, 1–24 (2008). source
Not checked in the free run.
- cites [15] Locke, E. A. & Latham, G. P. A Theory of Goal Setting and Task Performance (Prentice-Hall Inc., Michigan, 1990).
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable).
PremisecitationPersonalization is especially important for effective learning in order to cater to students with different backgrounds.
“In order to cater to students with different backgrounds, personalization is especially important for effective learning.6”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [6] Savi, A. O., van der Maas, H. L. J. & Maris, G. K. J. Open learning at a distance: lessons for struggling MOOCs. Science 347, 958 (2015). source
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable), S4 (based).
InterpretationreasoningFor courses on subjects with clearly defined dependence among topics, letting all students follow the same learning path is advantageous.
“For courses on subjects with clearly defined dependence among the topics (e.g. normally learning to perform addition comes before learning to perform subtraction in mathematics), letting all students follow the same learning path is advantageous.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
InterpretationreasoningFor courses with unclear dependencies, it may be advantageous to allow students more freedom to access learning contents without strictly following the pre-defined sequence based on personal situations.
“For these types of courses, it may be advantageous to allow students more freedom to access the learning contents without strictly following the pre-defined sequence based on their personal situations (e.g. background knowledge, time constraints).”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
PremisecitationGoal Net is an AI knowledge representation tool designed to model logic flows which can guide actions of software agents in a visual way without writing lengthy programme codes.
“It is an AI knowledge representation tool designed to model logic flows which can guide the actions of software agents in a visual way without having to write lengthy programme codes.14”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [13] Yu, H., Shen, Z. & Miao, C. A goal oriented development tool to automate the incorporation of intelligent agents into interactive digital media applications. ACM Comput. Entertain. 6, 1–24 (2008). source
Not checked in the free run.
- cites [14] Shen, Z., Yu, H., Miao, C., Li, S. & Chen, Y. Multi-agent system development MADE easy. In AAAI, 4391–4392 (2016). source
Not checked in the free run.
Probably not a substantial claim: may not pass S4 (based).
Interpretationnone givenCurrent MOOC teaching is still more focused on standardization, not personalization.
“The main reason is likely to be that current MOOC teaching is still more focused on standardization, not personalization.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationIn crowdsourcing, data-driven algorithmic management based on stochastic optimization have been used to dynamically divide work, fairly allocate workload, achieve superlinear collective productivity, schedule rest-breaks, and adjust incentives.
“In the domain of crowdsourcing, data-driven algorithmic management approaches based on stochastic optimization have been used to dynamically divide work among workers so as to fairly allocation workload and achieve superlinear collective productivity,30–34 schedule rest-breaks35 and adjust incentives to encourage participation based on game theory.36,37”
Quote in the paperSubstantial claim0 of 8 sources checked- cites [30] Yu, H., Miao, C., An, B., Leung, C. & Lesser, V. R. A reputation management model for resource constrained trustee agents. In IJCAI, 418–424 (2013).
Not checked in the free run.
- cites [31] Yu, H. et al. A multi-agent game for studying human decision-making. In AAMAS, 1661–1662 (2014).
Not checked in the free run.
- cites [32] Yu, H. et al. Efficient task sub-delegation for crowdsourcing. In AAAI, 1305–1311 (2015). source
Not checked in the free run.
- cites [33] Yu, H. et al. Mitigating herding in hierarchical crowdsourcing networks. Sci. Rep. 6, 4 (2016). source
Not checked in the free run.
- cites [34] Yu, H. et al. Algorithmic management for improving collective productivity in crowdsourcing. Sci. Rep. 7, 12541 (2017). source
Not checked in the free run.
- cites [35] Yu, H., Shen, Z., Fauvel, S. & Cui, L. Efficient scheduling in crowdsourcing based on workers’ mood. In ICA, 121–126 (2017). source
Not checked in the free run.
- cites [36] Tran-Thanh, L., Venanzi, M., Rogers, A. & Jennings, N. R. Efficient budget allocation with accuracy guarantees for crowdsourcing classification tasks. In AAMAS, 901–908 (2013).
Not checked in the free run.
- cites [37] Biswas, A., Jain, S., Mandal, D. & Narahari, Y. A truthful budget feasible multi-armed bandit mechanism for crowdsourcing time critical tasks. In AAMAS, 1101–1109 (2015).
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
Interpretationnone givenLearning outcomes are often hard to define and hard to measure.
“One of the most important is that learning outcomes are often hard to define and hard to measure (e.g. are a high test scores indicative of good learning outcomes?).”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based).
InterpretationreasoningUsing a combination of micro-MOOCs and learning path construction tools, instructors can specify a general syllabus with improved efficiency and flexibility of updating.
“Using a combination of micro-MOOCs and learning path construction tools (possibly derived from Goal Net or other similar tools such as the Belief-Desire-Intention (BDI) model16), instructors can specify a general syllabus consisting of finely grained learning contents and activities with improved efficiency and flexibility of updating.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationJill Watson is helping the Knowledge Based Artificial Intelligence MOOC to answer students' questions.
“Currently, Jill Watson,39 an AI teaching assistant which is built on top of the IBM Watson platform, is helping the Knowledge Based Artificial Intelligence MOOC offered by Georgia Institute of Technology to answer students’ questions.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [39] Goel, A. K. & Joyner, D. A. Using AI to teach AI: lessons from an online AI class. AI Mag. 38, 48–58 (2017). source
Not checked in the free run.
Classified as probably not a substantial claim. May combine two claims.
InterpretationreasoningVLCs may be designed to help learners adopt more effective learning habits, and such applications might be perceived as violating learners' personal autonomy.
“As VLCs may be designed to help learners adopt more effective learning habits through learning skill analytics and smart interventions, such applications might be perceived as violating learners’ personal autonomy.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationA remembrance companion was proposed to help learners deal with large volumes of knowledge.
“A remembrance companion was proposed to help learners deal with large volumes of knowledge.42”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [42] Wu, Q., Han, X., Yu, H., Miao, C. & Shen, Z. The innovative applications of learning companions in Virtual Singapura. In AAMAS, 1171–1172 (2013).
Not checked in the free run.
May combine two claims.
PremisecitationCurrently, community TAs are recruited on a small scale (typically less than 10 per course) and mainly perform discussion moderation tasks.
“Currently, they are recruited on a small scale (typically less than 10 community TAs per course) and mainly perform discussion moderation tasks.21”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [21] Papadopoulos, K., Sritanyaratana, L. & Klemmer, S. R. Community TAs scale high-touch learning, provide student-staff brokering, and build esprit de corps. In L@S, 163–164 (2014). source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Interpretationnone givenAs software agents, VLCs can address the problem of large-scale usage in MOOCs.
“As software agents, VLCs can address the problem of large-scale usage in MOOCs.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationGoal Net enables transitions among goals and achievement of goals to be tracked.
“It also enables transitions among goals and the achievement of goals to be tracked.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [13] Yu, H., Shen, Z. & Miao, C. A goal oriented development tool to automate the incorporation of intelligent agents into interactive digital media applications. ACM Comput. Entertain. 6, 1–24 (2008). source
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable), S4 (based).
InterpretationreasoningVLCs need to be able to hold meaningful conversations to effectively facilitate learning of complex topics.
“Firstly, in order to effectively facilitate learning of complex topics instead of just answering questions, VLCs need to be able to hold meaningful conversations (either vocally or textually).”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationBetty's Brain has been shown to help learners consolidate their knowledge.
“A teachable learning companion—Betty’s Brain—has been shown to help learners consolidate their knowledge.43”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [43] Biswas, G., Segedy, J. R. & Leelawang, K. From design to implementation to practice—a learning by teaching system: Betty’s brain. Int. J. Artif. Intel. Educ. 26, 350–364 (2016). source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationA curious companion was proposed to provide personalized guidance to learners in a virtual world-based learning environment.
“A curious companion was proposed40,41 to provide personalized guidance to learners in a virtual world-based learning environment.”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [40] Wu, Q. & Miao, C. Modeling curiosity-related emotions for virtual peer learners. IEEE Comput. Intell. Mag. 8, 50–62 (2013). source
Not checked in the free run.
- cites [41] Wu, Q., Liu, S., Miao, C., Liu, Y. & Leung, C. A social curiosity inspired recommendation model to improve precision, coverage and diversity. In WI, 240–247 (2016). source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
InterpretationreasoningAI models trained on available learning behaviour data may be biased towards high performing learners.
“In addition, as the available learning behaviour data are mostly from people who managed to complete the MOOCs, it is possible that the AI models trained from such data may be biased towards high performing learners.”
Quote in the paperNot a substantial claim- Rests on an assumption: Available learning behaviour data are mostly from people who managed to complete the MOOCs.
Probably not a substantial claim: may not pass S2 (checkable), S4 (based).
Interpretationnone givenCurrent MOOC platform infrastructure is designed to efficiently help students 'learn facts' rather than 'acquire skills'.
“It can be argued that the current MOOC platform infrastructure is designed to efficiently help students 'learn facts' rather than 'acquire skills'.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationnone givenRunning interactive sessions require time and effort which becomes unfeasible for MOOCs with large student populations.
“However, running such interactive sessions require time and effort which becomes unfeasible for MOOCs with large student populations.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
Interpretationnone givenIf MOOC platform operators incorporate support for personalization, richer interaction opportunities and analytics with proper ethical oversight, MOOCs can be transformed into test-beds for advancing educational research, and ultimately, improve learning.
“If MOOC platform operators incorporate support for personalization of learning, richer interaction opportunities and analytics of the resulting learning behaviour data into their system framework with proper ethical oversight mechanisms, MOOCs can be transformed into test-beds for advancing educational research, and ultimately, improve learning.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).
Interpretationnone givenIn order to provide believable interactions, VLCs need to be equipped with human-like characteristics.
“In order to provide believable interactions, VLCs need to be equipped with human-like characteristics.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationnone givenInteractions with VLCs can generate learning behaviour data to facilitate learning skill analytics.
“Interactions with VLCs can also generate learning behaviour data to facilitate learning skill analytics.”
Quote in the paperNot a substantial claimMay combine two claims.
PremisecitationRubrics derived from learners' performance in learning tasks reflect students' learning skills to some extent, but are incomplete.
“Rubrics23 derived from learners’ performance in learning tasks reflect students’ learning skills to some extent, but are incomplete.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [23] Stevens, D. D. & Levi, A. J. Introduction to Rubrics: An Assessment Tool to Save Grading Time, Convey Effective Feedback, and Promote Student Learning (Stylus Publishing, LLC, Sterling, Virginia, 2013).
Not checked in the free run.
The paper relies on this without support.
Available learning behaviour data are mostly from people who managed to complete the MOOCs.
“as the available learning behaviour data are mostly from people who managed to complete the MOOCs”
The paper relies on this without support.
In MOOCs, community TAs often participate on a volunteer basis.
“In MOOCs, community TAs often participate on a volunteer basis.”
The paper relies on this without support.
The choice of incorporating interaction techniques into a MOOC is up to the course instructor, so the use of these techniques across courses varies.
“In addition, as the choice of incorporating these interaction techniques into a MOOC is up to the course instructor, the use of these techniques across courses varies.”
The paper relies on this without support.
Teaching machines to converse like humans turns out to be much harder than originally anticipated.
“However, teaching machines to converse like humans turns out to be much harder than originally anticipated.”
The paper relies on this without support.
AI conversational interaction is still an open research challenge today.
“AI conversational interaction is still an open research challenge today.”
Cites
Pan, S. J., & Yang, Q. (2010) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2009.191 · Cited by 23,387 · see citing works
GAN（Generative Adversarial Nets）
淳司, 柴. (2017) · Journal of Japan Society for Fuzzy Theory and Intelligent Informatics · doi:10.3156/jsoft.29.5_177_2 · Cited by 19,268 · see citing works
Weinberger, A., & Fischer, F. (2005) · Computers & Education · doi:10.1016/j.compedu.2005.04.003 · Cited by 902 · see citing works
Massive open online course completion rates revisited: Assessment, length and attrition
Jordan, K. (2015) · The International Review of Research in Open and Distributed Learning · doi:10.19173/irrodl.v16i3.2112 · Cited by 510 · see citing works
Stevens, D. D., & Levi, A. J. (2004) · Cited by 469 · see citing works
Peer and self assessment in massive online classes
Kulkarni, C., Wei, K. P., Le, H., Chia, D., Papadopoulos, K., Cheng, J., Koller, D., & Klemmer, S. R. (2013) · ACM Transactions on Computer-Human Interaction · doi:10.1145/2505057 · Cited by 390 · cited 1× · see citing works
“Techniques such as involvement of community teaching assistants (TAs), 21 peer assessment tools, 22 well-designed elaborated rubrics with speci fi c content feedback 23 and video feedbacks 24 are available to help MOOC instructors scale up the provision of perceived personal attention to students.”
Method(2008) · Cited by 365 · see citing works
Reich, J. (2015) · Science · doi:10.1126/science.1261627 · Cited by 294 · see citing works
Redefining the learning companion: the past, present, and future of educational agents
Chou, C., Chan, T.-W., & Lin, C.-J. (2003) · Computers & Education · doi:10.1016/s0360-1315(02)00130-6 · Cited by 270 · see citing works
Marshall, P., Antle, A. N., van den Hoven, E., & Rogers, Y. (2013) · ACM Transactions on Computer-Human Interaction · doi:10.1145/2442106.2442107 · Cited by 231 · see citing works
Identifying At-Risk Students in Massive Open Online Courses
He, J., Bailey, J., Rubinstein, B. I. P., & Zhang, R. (2015) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v29i1.9471 · Cited by 215 · see citing works
Using technology in Higher Education: an Activity Theory perspective
Issroff, K., & Eileen Scanlon, A. (2002) · Journal of Computer Assisted Learning · doi:10.1046/j.0266-4909.2001.00213.x · Cited by 179 · see citing works
Mitigating Herding in Hierarchical Crowdsourcing Networks
Yu, H., Miao, C., Leung, C., Chen, Y., Fauvel, S., Lesser, V., & Yang, Q. (2016) · Scientific Reports · doi:10.1038/s41598-016-0011-6 · Cited by 171 · see citing works
Self-citationFrom Design to Implementation to Practice a Learning by Teaching System: Betty’s Brain
Biswas, G., Segedy, J. R., & Bunchongchit, K. (2015) · International Journal of Artificial Intelligence in Education · doi:10.1007/s40593-015-0057-9 · Cited by 167 · see citing works
Moral Decision Making Frameworks for Artificial Intelligence
Conitzer, V., Sinnott-Armstrong, W. P., Borg, J. S., Deng, Y. Q., & Kramer, M. (2017) · AAAI Publications (The Association for the Advancement of Artificial Intelligence (AAAI)) · doi:10.1609/aaai.v31i1.11140 · Cited by 133 · see citing works
Education online: The virtual lab
Waldrop, M. M. (2013) · Nature · doi:10.1038/499268a · Cited by 114 · see citing works
Student skill and goal achievement in the mapping with google MOOC
Wilkowski, J., Deutsch, A., & Russell, D. M. (2014) · Proceedings of the first ACM conference on Learning @ scale conference · doi:10.1145/2556325.2566240 · Cited by 89 · see citing works
Humans display a reduced set of consistent behavioral phenotypes in dyadic games
Poncela-Casasnovas, J., Gutiérrez-Roig, M., Gracia-Lázaro, C., Vicens, J., Gómez‐Gardeñes, J., Perelló, J., Moreno, Y., Duch, J., & Sánchez, Á. (2016) · Science Advances · doi:10.1126/sciadv.1600451 · Cited by 83 · see citing works
Efficient budget allocation with accuracy guarantees for crowdsourcing classification tasks
Tran-Thanh, L., Venanzi, M., Rogers, A., et al. (2013) · Adaptive Agents and Multi-Agents Systems · doi:10.65109/vown7123 · Cited by 83 · see citing works
Methodological Approach and Technological Framework to Break the Current Limitations of MOOC Model
Fidalgo-Blanco, Á., Séin-Echaluce, M. L., & García‐Peñalvo, F. J. (2015) · LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas) · Cited by 82 · see citing works
On Computing Explanations in Argumentation
Fan, X., & Toni, F. (2015) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v29i1.9420 · Cited by 79 · see citing works
Using AI to Teach AI: Lessons from an Online AI Class
Goel, A. K., & Joyner, D. A. (2017) · AI Magazine · doi:10.1609/aimag.v38i2.2732 · Cited by 75 · see citing works
Open Learning at a Distance: Lessons for Struggling MOOCs
McAndrew, P., & Scanlon, E. (2013) · Science · doi:10.1126/science.1239686 · Cited by 68 · see citing works
BDI Agent Programming in AgentSpeak Using Jason (Tutorial Paper)
Bordini, R. H., & Hübner, J. (2005) · Computational Logic in Multi-Agent Systems · doi:10.1007/11750734_9 · Cited by 67 · see citing works
Backward Design: Targeting Depth of Understanding for All Learners
Childre, A., Sands, J. R., & Pope, S. T. (2009) · TEACHING Exceptional Children · doi:10.1177/004005990904100501 · Cited by 62 · see citing works
Algorithmic Management for Improving Collective Productivity in Crowdsourcing
Yu, H., Miao, C., Chen, Y., et al. (2017) · Scientific Reports · doi:10.1038/s41598-017-12757-x · Cited by 54 · see citing works
Self-citationCorrelating skill and improvement in 2 MOOCs with a student's time on tasks
Champaign, J., Colvin, K. F., Liu, A., Fredericks, C., Seaton, D., & Pritchard, D. E. (2014) · Proceedings of the first ACM conference on Learning @ scale conference · doi:10.1145/2556325.2566250 · Cited by 50 · see citing works
Gillani, N., Yasseri, T., Eynon, R., et al. (2014) · Scientific Reports · doi:10.1038/srep06447 · Cited by 48 · see citing works
Scale-Driven Automatic Hint Generation for Coding Style
Choudhury, R. R., Yin, H., & Fox, A. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-39583-8_12 · Cited by 43 · see citing works
Efficient Task Sub-Delegation for Crowdsourcing
Yu, H., Miao, C., Shen, Z., et al. (2015) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v29i1.9337 · Cited by 42 · see citing works
Self-citationA reputation management approach for resource constrained trustee agents
Yu, H., Miao, C., An, B., et al. (2013) · International Joint Conference on Artificial Intelligence · Cited by 40 · see citing works
Self-citationA Truthful Budget Feasible Multi-Armed Bandit Mechanism for Crowdsourcing Time Critical Tasks
Biswas, A., Jain, S., Mandal, D., et al. (2015) · Adaptive Agents and Multi-Agents Systems · doi:10.5555/2772879.2773291 · Cited by 38 · see citing works
Yu, H., Shen, Z., & Miao, C. (2008) · Computers in entertainment · doi:10.1145/1371216.1371227 · Cited by 34 · see citing works
Self-citationThe innovative application of learning companions in virtual singapura
Wu, Q., Han, X., Yu, H., Shen, Z., & Miao, C. (2013) · Adaptive Agents and Multi-Agents Systems · doi:10.5555/2484920.2485126 · Cited by 32 · see citing works
Self-citationA multi-agent game for studying human decision-making
Yu, H., Yu, X., Lim, S. F., et al. (2014) · Adaptive Agents and Multi-Agents Systems · doi:10.5555/2615731.2616113 · Cited by 32 · see citing works
Self-citationModeling Curiosity-Related Emotions for Virtual Peer Learners
Wu, Q., & Miao, C. (2013) · IEEE Computational Intelligence Magazine · doi:10.1109/mci.2013.2247826 · Cited by 24 · see citing works
Self-citationKrafft, P. M., Macy, M. W., & Pentland, A. (2017) · ACM Conference on Computer Supported Cooperative Work (CSCW) · doi:10.1145/2998181.2998354 · Cited by 24 · see citing works
A Social Curiosity Inspired Recommendation Model to Improve Precision, Coverage and Diversity
Wu, Q., Liu, S., Miao, C., Liu, Y., & Leung, C. (2016) · IEEE/WIC/ACM International Conference on Web Intelligence · doi:10.1109/wi.2016.0042 · Cited by 20 · see citing works
Self-citationBots as Virtual Confederates: Design and EthicsPreprint
Krafft, P. M., Macy, M. W., & Pentland, A. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1611.00447 · Cited by 19 · see citing works
A dataset of human decision-making in teamwork management
Yu, H., Shen, Z., Miao, C., Leung, C., Chen, Y., Fauvel, S., Lin, J., Cui, L., Pan, Z., & Yang, Q. (2017) · Scientific Data · doi:10.1038/sdata.2016.127 · Cited by 18 · see citing works
Self-citationEfficient scheduling in crowdsourcing based on workers' mood
Yu, H., Shen, Z., Fauvel, S., et al. (2017) · IEEE International Conference on Agents (ICA) · doi:10.1109/agents.2017.8015317 · Cited by 15 · see citing works
Self-citationNavigating Massive Open Online Courses
Savi, A. O., van der Maas, H. L. J., & Maris, G. (2015) · Science · doi:10.1126/science.347.6225.958 · Cited by 12 · see citing works
Multi-Agent System Development MADE Easy
Shen, Z., Yu, H., Miao, C., et al. (2016) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v30i1.9827 · Cited by 12 · see citing works
Self-citationVorobeychik, Y., An, B., & Tambe, M. (2012) · Adaptive Agents and Multi-Agents Systems · doi:10.5555/2343896.2343977 · Cited by 10 · see citing works
MOOCs Meet Measurement Theory: A Topic-Modelling Approach
He, J., Rubinstein, B., Bailey, J., Zhang, R., Milligan, S., & Chan, J. (2016) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v30i1.10151 · Cited by 10 · see citing works
Community TAs scale high-touch learning, provide student-staff brokering, and build esprit de corps
Papadopoulos, K., Sritanyaratana, L., & Klemmer, S. R. (2014) · Proceedings of the first ACM conference on Learning @ scale conference · doi:10.1145/2556325.2567860 · Cited by 7 · see citing works
Losing the Red Pen: Video Grading Feedback in Distance and Blended Learning Writing Courses.
Jones, L. A. (2014) · Cited by 2 · see citing works
Flagged first, then by citations
Research works printed but not linked: 2
- Pellegrino, J. W. & Hilton, M. L. (eds.). Education for Life and Work: Developing Transferable Knowledge and Skills in the 21st Century (National Academies Press, Washington DC, 2012).
- Goodfellow, I. et al. Generative adversarial nets. In Advances in Neural Information Processing Systems Vol. 27 (eds. Ghahramani, Z. et al.) 2672–2680 (2014).
Cited by
Sorted by citations
Digital twin: values, challenges and enablers from a modeling perspective
Rasheed, A., San, O., & Kvamsdal, T. (2020) · IEEE Access · doi:10.1109/access.2020.2970143 · Cited by 1,952 · see citing works
Artificial Intelligence in Education (AIEd): a high-level academic and industry note 2021
Chaudhry, M. A., & Kazim, E. (2022) · AI and Ethics · doi:10.1007/s43681-021-00074-z · Cited by 425 · see citing works
Generative AI in Higher Education: Balancing Innovation and Integrity
Francis, N. J., Jones, S., & Smith, D. P. (2025) · British Journal of Biomedical Science · doi:10.3389/bjbs.2024.14048 · Cited by 295 · see citing works
Building Ethics into Artificial Intelligence
Yu, H., Shen, Z., Miao, C., Leung, C., Lesser, V. R., & Yang, Q. (2018) · Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2018/779 · Cited by 210 · see citing works
Self-citationFrom concerns to benefits: A comprehensive study of ChatGPT usage in education
Jo, H. (2024) · International Journal of Educational Technology in Higher Education · doi:10.1186/s41239-024-00471-4 · Cited by 146 · see citing works
Artificial intelligence in education and assessment methods
Braiki, B. A., Harous, S., Zaki, N., & Alnajjar, F. (2020) · Bulletin of Electrical Engineering and Informatics · doi:10.11591/eei.v9i5.1984 · Cited by 79 · see citing works
- Authors
- Yu, H., Miao, C., Leung, C., & White, T. J.
- Affiliations
- Nanyang Technological University (SG)
- University of British Columbia (CA)
- Corresponding author
- Yu, Han
- Miao, Chunyan
- Venue
- npj Science of Learning
- Published
- 2017-12-14
- Volume
- 2
- Issue
- 1
- Pages
- 15-15
- ISSN
- 2056-7936 (ISSN-L)
- PMC
- PMC6220236
- OpenAlex
- W2769019609
- Semantic Scholar
- 9e1289df0e1177662bbe066ac9b6ffa8450dee0d
- Cited by
- 131
- References
- 47
- Citations by year
- 2026: 10 · 2025: 18 · 2024: 18 · 2023: 23 · 2022: 17 · 2021: 20 · 2020: 11 · 2019: 9 · 2018: 5
- FWCI
- 1.56
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- npj Science of Learning (publishedVersion, cc-by)
- Europe PMC (PubMed Central) (publishedVersion, other-oa)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
- DR-NTU (Nanyang Technological University) (submittedVersion, cc-by)
- Funders
- National Research Foundation
- Nanyang Technological University
- Peking University
- Ng Teng Fong Charitable Foundation
- Topics
- Physical Sciences › Computer Science › Computer Science Applications › Online Learning and Analytics
- Physical Sciences › Computer Science › Information Systems › Artificial Intelligence in Education
- Physical Sciences › Computer Science › Artificial Intelligence › Intelligent Tutoring Systems and Adaptive Learning
- Keywords
- MOOCs, artificial intelligence in education, personalized learning, knowledge representation, adaptive learning paths, optimization, emotional agents, learning analytics, AI ethics in education
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
