Pith. sign in

REVIEW 2 major objections 6 minor 74 references

ParaTutor: Coordinating Parent and Child Math Tutoring through Role Separated LLM Scaffolding

T0 review · 2 major / 6 minor · reviewed 2026-07-15 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read In family math tutoring, LLM value depends on splitting support by role—not only on model power.

desk verdict Useful HCI system paper on role-separated LLM scaffolding for parent–child math tutoring; the coordination claim is directionally right but not yet cleanly isolated from visuals and UI structure. read the letter →

arxiv 2606.18030 v2 pith:QHMPSNLQ submitted 2026-06-16 cs.HC

Lan Luo , Anqi Wang , Muzhi Zhou , Junhua Zhu , Jie Cai , Ao Yu , Hui Pan This is my paper · ORCID
classification cs.HC
keywords LargeLanguageModelsParentalInvolvementMathematicalWordProblemsScaffoldingHuman-ComputerInteractionFamilyLearningRole-SeparatedInterfaces
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Parent–child math tutoring is asymmetric: parents guide, children must stay in the reasoning seat. Most LLM learning tools ignore that split and treat users as a single learner or a symmetric pair. A formative study of Chinese home tutoring found recurring breakdowns—parents who cannot draw relations, knowledge and method mismatches with school teaching, and communication that escalates into conflict. ParaTutor answers with role-separated, phase-gated scaffolding: parents receive strategy, language, repair, and phase prompts; children receive visual grounding without solution steps; phase advance needs joint confirmation. In 23 dyads comparing no AI, generic conversational LLM help, and two ParaTutor modes, generic chat often supplied explanations yet reduced the parent’s instructional role, while role-separated support redistributed tutoring work, kept children engaged with word problems, and helped parents turn model methods into child-facing moves. The paper’s claim is that in multi-user family learning, how support is coordinated across roles matters as much as raw model capability.

What carries the argument

ParaTutor: a multi-agent, phase-gated system with role-separated interfaces—parent-facing strategy/language/repair scaffolds across understanding–calculation–summarization, child-facing visual grounding generated from structured problem state without procedural answers, and joint confirmation before phase advance.

What would settle it

Hold visuals and interface quality fixed and remove role separation (same content as shared chat or child-facing tutor only): if parent instructional involvement and child reasoning engagement then match ParaTutor, the claim that role-coordinated delivery is the active ingredient fails.

Watch

Extended reading notes

Core claim

Generic conversational LLM assistance can give useful math explanations yet still sideline parents and undercut children’s active reasoning; distributing support by role—strategic scaffolds to parents, visual grounding to children, with phase-gated shared progression—better preserves parent-led tutoring and children’s participation in problem solving.

Load-bearing premise

Short-term reports from 23 Chinese parent–child dyads, plus one time comparison of aligned versus complementary strategies, are enough to pin the benefit on role separation rather than diagrams, interface polish, or novelty alone.

Editorial extensions

If this is right

  • Family LLM tutors should route different surfaces to parents and children instead of one shared chat that can become the primary instructor.
  • Phase-gated progression with joint confirmation can slow answer-jumping and keep children responsible for intermediate reasoning.
  • Scaffolds aligned with parents’ existing habits speed real-time coordination; complementary strategies expand solution paths but raise live cognitive load.
  • Parent-facing language scaffolds (neutral rephrasings, positive starters) can buffer emotional escalation during correction.
  • Evaluations of home educational LLMs should measure role preservation and child reasoning engagement, not only explanation quality.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same role-separation pattern may transfer to other asymmetric guide–learner dyads where one person must remain the instructor rather than a co-user of a shared tutor.
  • Longer use might show whether parents internalize scaffolds and need less system support over time, or whether burden stays high for low-confidence parents.
  • A factorial control isolating visual grounding from role-separated parent prompts would show which design lever mainly drives children’s engagement versus parents staying in the loop.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 6 minor

Summary. The paper presents ParaTutor, a multi-agent LLM system for parent–child math word-problem tutoring that separates support by role: parents receive phase-gated strategy, language, and repair scaffolds, while children receive visual grounding without procedural solutions. A formative study with 11 parents and 2 teachers identifies role-maintenance practices and three recurring breakdowns (visualization difficulty, knowledge/method gaps, communication conflict). A within-subjects Latin-square evaluation with 23 dyads (children 10–12) compares conventional tutoring, generic conversational LLM (DeepSeek), and two ParaTutor variants (aligned vs complementary strategies). Interview and limited quantitative results are used to argue that generic LLM help tends to sideline parents, whereas role-separated, phase-gated scaffolding better preserves parent-led tutoring and children’s active reasoning, implying that LLM value in family learning depends on coordination across asymmetric roles.

Significance. If the central claim holds, the work supplies a concrete design pattern—role-separated surfaces plus phase-gated orchestration—for LLM systems in asymmetric multi-user learning, an underexplored setting relative to single-user tutors or symmetric collaboration. Strengths include a formative-to-system pipeline grounded in local tutoring practice, an explicit dual-interface architecture with answer-disclosure constraints, a four-condition Latin-square design, and candid discussion of parental burden and transfer limits. These elements make the paper a useful HCI contribution even if causal isolation of role separation remains incomplete.

major comments (2)
  1. §5.1–5.3: The central claim that coordination (role separation + phase gating), not model capability or interface polish, drives parent-led tutoring is not yet secured by the design. Mode B is open conversational LLM access for either user; Modes C/D add dual panels, child-facing visual grounding without procedures, shared phase confirmation, and strategy language. Reported advantages rest mainly on post-session interview quotes plus one paired t-test on completion time between aligned and complementary ParaTutor modes (328 s vs 365 s). There is no process coding of turn-taking, parental question rate, answer-disclosure events, or independent learning outcomes, and no control that holds visuals/structure constant while removing role separation (or vice versa). Without such isolation or behavioral measures, diagrams, phase UI, or novelty remain plausible alternative explanations for the r
  2. §5.3–5.5 and abstract: Quantitative support for the comparative claim is thin relative to the strength of the conclusions. Beyond the single time comparison and SUS item means, the paper does not report accuracy, strategy transfer, engagement counts, or inter-rater reliability for the thematic analysis. The abstract and discussion assert that ParaTutor “helped redistribute tutoring work,” “increased children’s engagement,” and “better preserves parent-led support”; these should be tempered or backed by additional coded interaction metrics before the coordination claim is treated as established.
minor comments (6)
  1. Title/abstract vs body: the arXiv title and abstract wording differ slightly from the manuscript title and abstract (e.g., “Coordinating… Role Separated” vs “LLM-Mediated… Role-Separated Scaffolding Interface”); align them.
  2. §4.3 / Fig. 2: the rule-based state-to-diagram pipeline and phase-completion criteria are described at a high level; a short appendix example of entity–relation extraction to diagram would improve reproducibility.
  3. §5.1: clarify whether children could query DeepSeek directly in Mode B and how often that occurred; this affects interpretation of “parent role reduction.”
  4. Fig. 3: SUS boxplot is useful but item wording and scale anchors should be stated in the caption or text for readers unfamiliar with the instrument.
  5. Placeholder ACM metadata (Conference acronym ’XX, Woodstock NY, 2018) should be cleaned for the camera-ready version.
  6. Related work is solid on scaffolding and family AI; a brief note on prior dual-user or parent-facing math tools would further situate the contribution.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical formative-to-design-to-evaluation chain with independent user-study outcomes, not definitional or fitted-by-construction results.

full rationale

ParaTutor is an HCI systems paper. Design requirements R1–R3 are derived from formative challenges C1–C3 (visualization gaps, knowledge gaps, communication conflict) identified via thematic analysis of 11 parents + 2 teachers; the multi-agent system and dual-panel interface implement those requirements; a separate within-subjects evaluation with 23 new dyads then compares four modes (no-AI, generic DeepSeek, ParaTutor-aligned, ParaTutor-complementary) via interviews, SUS, and one paired t-test on completion time. Success metrics (parents remaining in the instructional loop, children’s engagement, reduced tension) are not guaranteed by construction of the interface or by any fitted parameter renamed as a prediction. Mode C being “aligned” to strategies elicited in the formative study is an experimental factor, not a tautology: the paper still measures and reports empirical differences (e.g., 328 s vs 365 s) and trade-offs. There are no equations, no uniqueness theorems imported from the authors’ prior work, no ansatz smuggled via self-citation, and no renaming of a known result presented as novel derivation. The evaluation’s causal isolation weaknesses (confounds of visuals/structure with role separation) are validity concerns, not circularity. The derivation chain is therefore self-contained and non-circular.

Assumptions & free parameters 2 free parameters · 4 assumptions · 2 invented entities

This is an HCI system paper. The central claim rests on standard scaffolding theory, a formative-derived role model, and empirical observations from a small convenience sample rather than free parameters or invented physical entities. The main unproven loads are domain assumptions about how Chinese upper-elementary home tutoring works and that short-term self-report plus one timing test generalize.

free parameters (2)
  • session_time_limit_minutes
    Each tutoring session capped at 30 minutes with fixed breaks; affects observed completion times and strategy feasibility but is a protocol choice, not a fitted scientific constant.
  • phase_completion_criteria
    Rule-based phase transitions depend on predefined confirmation criteria over shared state; exact thresholds are design choices that gate when support changes.
assumptions (4)
  • domain assumption Effective parent-child math tutoring requires maintaining distinct asymmetric roles: parent guides process, child remains responsible for reasoning.
    Stated as formative finding and design foundation in §3 and §4.1; treated as normative rather than experimentally proven universal.
  • domain assumption Math word problem solving is usefully segmented into understanding, calculation, and summarization phases (Polya-style).
    Used to structure agents and UI phase gates (§3.3.1, §4.2); classical but not the only possible segmentation.
  • standard math Scaffolding should regulate timing and amount of support without giving final answers (Vygotsky/Rogoff tradition).
    Background educational theory cited in related work and design; standard in the domain though operationalization is paper-specific.
  • ad hoc to paper Withholding procedural solutions from the child-facing surface while giving strategy language to parents will preserve parent agency better than open chat.
    Core design hypothesis of ParaTutor (§4); tested only via qualitative comparison, not isolated experimentally.
invented entities (2)
  • ParaTutor multi-agent phase-gated orchestrator with role-separated parent/child surfaces
    purpose: Coordinate strategy scaffolds to parents and visual grounding to children across tutoring phases
    System artifact introduced by the paper; independent evidence is the user study, not external measurement.
  • Rule-based state-to-diagram visual grounding module
    purpose: Externalize entities/relations for children without revealing solution procedures
    Implementation component; no external validation of diagram correctness or learning effect separate from the whole system.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ParaTutor: Coordinating Parent and Child Math Tutoring through Role Separated LLM Scaffolding." pith.science (2026). https://pith.science/paper/QHMPSNLQ

@misc{pith2026260618030,
  author       = {Pith},
  title        = {Pith review of: ParaTutor: Coordinating Parent and Child Math Tutoring through Role Separated LLM Scaffolding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QHMPSNLQ}},
  note         = {Machine review of arXiv:2606.18030}
}
read the original abstract

Parent and child tutoring is a collaborative learning setting with asymmetric roles. Parents guide children s problem solving, while children are expected to remain actively engaged in understanding and reasoning. However, most LLM based learning systems are designed for single users or relatively symmetric collaboration, leaving parent and child tutoring with distinct instructional roles underexplored. Through a formative study, we found that parent and child math tutoring was often disrupted by cognitive misalignment, emotional escalation, and method mismatch. To address these challenges, we present ParaTutor, a multiple agents LLM based scaffolding system for home math word problem tutoring. ParaTutor distributes support across user roles by providing parents with strategy, language, repair, and phase scaffolds, while providing children with visual grounding for problem interpretation. We evaluated ParaTutor with 23 parent and child dyads (children aged 10 to 12) across four tutoring conditions that varied how LLM assistance was delivered. Results show that generic LLM assistance often provided useful explanations but did not consistently support parent led tutoring or children s active reasoning. In contrast, ParaTutor helped redistribute tutoring work across parents and children, increased children s engagement with word problems, supported shared understanding through visual grounding, and helped parents translate LLM generated methods into child facing tutoring moves. These findings suggest that in family learning, the value of LLM support depends not only on model capability, but also on how support is coordinated across users with different roles. Our work contributes design implications for LLM systems that support role sensitive scaffolding in parent and child learning.

Figures

Figures reproduced from arXiv: 2606.18030 by the authors.

Figure 1
Figure 1. Dual interface of ParaTutor. The workspace is divided into a parent-facing panel (C) that provides strategy-based [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Implementation architecture of ParaTutor. A [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Perceived Usability of ParaTutor System: Box Plot with Item-Level Means 5.6 Limitations and Design Boundaries Keeping parents in the loop also preserves parental burden. A central goal of ParaTutor is to keep parents actively involved rather than allowing the model to take over tutoring. While this design helped sustain parent-child engagement, it also meant that parents still needed to interpret the system’s guidan… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Parent-Child Interaction Flowchart in LLM-Assisted [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 6
Figure 6. Figure 6: (d)(e) shows a participated child being faced with [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

74 extracted references · 4 linked inside Pith

  1. [1]

    Ananya Bhattacharjee, Yuchen Zeng, Sarah Yi Xu, Dana Kulzhabayeva, Minyi Ma, Rachel Kornfield, Syed Ishtiaque Ahmed, Alex Mariakakis, Mary P Czerwinski, Anastasia Kuzminykh, et al . 2024. Understanding the role of large language models in personalizing and scaffolding strategies to combat academic procrasti- nation. InProceedings of the 2024 CHI Conferenc...

    2024

  2. [2]

    Katerina Bodovski and George Farkas. 2007. Mathematics growth in early ele- mentary school: The roles of beginning knowledge, student engagement, and instruction.The Elementary School Journal108, 2 (2007), 115–130

    2007

  3. [3]

    Lisa Boonk, Hieronymus JM Gijselaers, Henk Ritzen, and Saskia Brand-Gruwel

  4. [4]

    A review of the relationship between parental involvement indicators and academic achievement.Educational research review24 (2018), 10–30

    2018

  5. [5]

    Jiaju Chen, Minglong Tang, Yuxuan Lu, Bingsheng Yao, Elissa Fan, Xiaojuan Ma, Ying Xu, Dakuo Wang, Yuling Sun, and Liang He. 2025. Characterizing LLM- Empowered Personalized Story Reading and Interaction for Children: Insights From Multi-Stakeholder Perspectives. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–24

    2025

  6. [6]

    Zixin Chen, Jiachen Wang, Yumeng Li, Haobo Li, Chuhan Shi, Rong Zhang, and Huamin Qu. 2025. CoGrader: Transforming Instructors’ Assessment of Project Reports through Collaborative LLM Integration. InProceedings of the 38th Annual ACM Symposium on User Interface Software and Technology. 1–18

    2025

  7. [7]

    Alan Y Cheng, Carolyn Q Zou, Anthony Xie, Matthew Hsu, Felicia Yan, Felicity Huang, David K Zhang, Arjun Sharma, Rashon Poole, Daniel Wan Rosli, et al

  8. [8]

    InProceedings of the 38th Annual ACM Symposium on User Interface Software and Technology

    Oak Story: Improving Learner Outcomes with LLM-Mediated Interactive Narratives. InProceedings of the 38th Annual ACM Symposium on User Interface Software and Technology. 1–17

Show all 74 references
  1. [9]

    I Want to Think Like an SLP

    Aayushi Dangol, Aaleyah Lewis, Hyewon Suh, Xuesi Hong, Hedda Meadan, James Fogarty, and Julie A Kientz. 2025. “I Want to Think Like an SLP”: A Design Exploration of AI-Supported Home Practice in Speech Therapy. InProceedings of the 2025 CHI Conference on Human Factors in Compu...

    2025

  2. [10]

    Paramveer S Dhillon, Somayeh Molaei, Jiaqi Li, Maximilian Golub, Shaochun Zheng, and Lionel Peter Robert. 2024. Shaping human-AI collaboration: Varied scaffolding levels in co-writing with language models. InProceedings of the 2024 CHI conference on human factors in computing ...

    2024

  3. [11]

    Vanessa Figueiredo. 2025. Designing smarter conversational agents for kids: Lessons from cognitive work and means-ends analyses.ACM Transactions on Computer-Human Interaction32, 6 (2025), 1–40

    2025

  4. [12]

    Marilyn Frankenstein. 2009. Developing a criticalmathematical numeracy through real real-life word problems. InWords and worlds. Brill, 111–130

    2009

  5. [13]

    Fengqiang Gao, Chunze Xu, Qing Lv, Yufei Zhao, and Lei Han. 2024. Parent-child communication and educational anxiety: a longitudinal analysis based on the common fate model.BMC Psychology12, 1 (2024), 594. https://doi.org/10.1186/ s40359-024-02093-x

    2024

  6. [14]

    Radhika Garg and Subhasree Sengupta. 2020. Conversational technologies for in-home learning: using co-design to understand children’s and parents’ perspec- tives. InProceedings of the 2020 CHI conference on human factors in computing systems. 1–13

    2020

  7. [15]

    Alyssa R Gonzalez-DeHass, Patricia P Willems, and Marie F Doan Holbein. 2005. Examining the relationship between parental involvement and student motiva- tion.Educational psychology review17, 2 (2005), 99–123

    2005

  8. [16]

    2025.Empowering Children’s AI Literacy Through Co-Creating Stories with LLM

    Ariel Han and Shenshen Han. 2025.Empowering Children’s AI Literacy Through Co-Creating Stories with LLM. Association for Computing Machinery, New York, NY, USA, 994–999. https://doi.org/10.1145/3713043.3731520

    2025 doi

  9. [17]

    Kunlei He, Julian Levine, Kelsyann Cervera, Santiago Ojeda-Ramirez, Ying Xu, and Mark Warschauer. 2024. A home study of parent-child co-reading with a bilingual conversational agent. InExtended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–8

    2024

  10. [18]

    Hui-Ru Ho, Nitigya Kargeti, Ziqi Liu, and Bilge Mutlu. 2025. SET-PAiREd: De- signing for Parental Involvement in Learning with an AI-Assisted Educational Robot.arXiv preprint arXiv:2502.17623(2025)

    2025 arXiv

  11. [19]

    Hui-Ru Ho, Nathan Thomas White, Edward M Hubbard, and Bilge Mutlu. 2023. Designing Parent-child-robot Interactions to Facilitate In-Home Parental Math Talk with Young Children. InProceedings of the 22nd Annual ACM Interaction Design and Children Conference. 355–366

    2023

  12. [20]

    Jiaxiong Hu, Junze Li, Yuhang Zeng, Dongjie Yang, Danxuan Liang, Helen Meng, and Xiaojuan Ma. 2024. Designing scaffolding strategies for conversational agents in dialog task of neurocognitive disorders screening. InProceedings of the 2024 CHI Conference on Human Factors in Com...

    2024

  13. [21]

    Julie Hui and Michelle L Sprouse. 2023. Lettersmith: Scaffolding written profes- sional communication among college students. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems. 1–17

    2023

  14. [22]

    Sandeep Kumar Jaiswal and Rashmi Choudhuri. 2017. A review of the relation- ship between parental involvement and students’ academic performance.The International Journal of Indian Psychology4, 3 (2017), 110–123

    2017

  15. [23]

    Zhanming Jie, Jierui Li, and Wei Lu. 2022. Learning to reason deductively: Math word problem solving as complex relation extraction.arXiv preprint Conference acronym ’XX, June 03–05, 2018, Woodstock, NY LUO Lan, WANG Anqi, ZHOU Muzhi, ZHU Junhua, CAI Jie, YU Ao, and PAN Hui ar...

    2022 arXiv

  16. [24]

    Hyoungwook Jin, Yoonsu Kim, Yeon Su Park, Bekzat Tilekbay, Jinho Son, and Juho Kim. 2024. Using Large Language Models To Diagnose Math Problem- solving Skills At Scale. InProceedings of the Eleventh ACM Conference on Learning @ Scale(Atlanta, GA, USA)(L@S ’24). Association for...

    2024 doi

  17. [25]

    Yongkwan Kim, Sang-Gyun An, Joon Hyub Lee, and Seok-Hyung Bae. 2018. Agile 3D sketching with air scaffolding. InProceedings of the 2018 CHI Conference on Human Factors in Computing Systems. 1–12

    2018

  18. [26]

    Walter Kintsch and James G Greeno. 1985. Understanding and solving word arithmetic problems.Psychological review92, 1 (1985), 109

    1985

  19. [27]

    Unggi Lee, Youngin Kim, Sangyun Lee, Jaehyeon Park, Jin Mun, Eunseo Lee, Hyeoncheol Kim, Cheolil Lim, and Yun Joo Yoo. 2024. Can we use GPT-4 as a mathematics evaluator in education?: Exploring the efficacy and limitation of LLM-based automatic assessment system for open-ended...

    2024

  20. [28]

    Yoonjoo Lee, John Joon Young Chung, Tae Soo Kim, Jean Y Song, and Juho Kim

  21. [29]

    InProceedings of the 2022 CHI Conference on Human Factors in Computing Systems

    Promptiverse: Scalable generation of scaffolding prompts through human- AI hybrid knowledge graph annotation. InProceedings of the 2022 CHI Conference on Human Factors in Computing Systems. 1–18

    2022

  22. [30]

    Shucheng Li, Lingfei Wu, Shiwei Feng, Fangli Xu, Fengyuan Xu, and Sheng Zhong. 2020. Graph-to-tree neural networks for learning structured input-output translation with applications to semantic parsing and math word problem.arXiv preprint arXiv:2004.13781(2020)

    2020 arXiv

  23. [31]

    Michael Xieyang Liu, Tongshuang Wu, Tianying Chen, Franklin Mingzhe Li, Aniket Kittur, and Brad A Myers. 2024. Selenite: Scaffolding online sensemaking with comprehensive overviews elicited from large language models. InProceed- ings of the 2024 CHI Conference on Human Factors...

    2024

  24. [32]

    Yujia Liu, Siyu Zha, Yuewen Zhang, Yanjin Wang, Yangming Zhang, Qi Xin, Lun Yiu Nie, Chao Zhang, and Yingqing Xu. 2025. BrickSmart: Leveraging Generative AI to Support Children’s Spatial Language Learning in Family Block Play. InProceedings of the 2025 CHI Conference on Human ...

    2025

  25. [33]

    Wenhan Lyu, Yimeng Wang, Tingting (Rachel) Chung, Yifan Sun, and Yixuan Zhang. 2024. Evaluating the Effectiveness of LLMs in Introductory Computer Science Education: A Semester-Long Field Study. InProceedings of the Eleventh ACM Conference on Learning @ Scale. ACM, 63–74. http...

    2024

  26. [34]

    Jiaju Ma, Chau Vu, Asya Lyubavina, Catherine Liu, and Jingyi Li. 2025. Compu- tational Scaffolding of Composition, Value, and Color for Disciplined Drawing. InProceedings of the 38th Annual ACM Symposium on User Interface Software and Technology. 1–15

    2025

  27. [35]

    Shuai Ma, Junling Wang, Yuanhao Zhang, Xiaojuan Ma, and April Yi Wang

  28. [36]

    InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems

    DBox: Scaffolding algorithmic programming learning through learner-LLM co-decomposition. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems. 1–20

    2025

  29. [37]

    Datius Titus Mutangira, Pambas Basil Tandika, and Geraldina Edward. 2024. Early grade pupils’ development of Mathematics skills: Perceived challenges and culturally relevant strategies for parental involvement.Educational Dimension 11 (2024), 220–243

    2024

  30. [38]

    Andre Nickow, Philip Oreopoulos, and Vincent Quan. 2020. The impressive effects of tutoring on prek-12 learning: A systematic review and meta-analysis of the experimental evidence. (2020)

    2020

  31. [39]

    Andre Joshua Nickow, Philip Oreopoulos, and Vincent Quan. 2020. The trans- formative potential of tutoring for PreK-12 learning outcomes: Lessons from randomized evaluations

    2020

  32. [40]

    Misato Nihei, Taiga Nohara, Ikuko Sugawara, and Takazumi Ono. 2024. Chatbots as Tools in Parent–Child Relationships. InInternational Conference on Human- Computer Interaction. Springer, 228–241

    2024

  33. [41]

    Benjamin D Nye, Arthur C Graesser, and Xiangen Hu. 2014. AutoTutor and family: A review of 17 years of natural language tutoring.International Journal of Artificial Intelligence in Education24 (2014), 427–469

    2014

  34. [42]

    Srishti Palani, Aakanksha Naik, Doug Downey, Amy X Zhang, Jonathan Bragg, and Joseph Chee Chang. 2023. Relatedly: Scaffolding literature reviews with existing related work sections. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems. 1–20

    2023

  35. [43]

    Melina Petsolari, Seray B Ibrahim, and Petr Slovak. 2024. Socio-technical imag- inaries: Envisioning and understanding AI parenting supports through design fiction. InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems. 1–27

    2024

  36. [44]

    George Polya. 2014. How to solve it: A new aspect of mathematical method. In How to solve it. Princeton university press

    2014

  37. [45]

    Yoppy Purnomo, Tri Ainun, Puteri Nina, Karunia Utami, Ririn Wijayanti, Siti Ismail, et al. 2022. Mother as a teacher at home: Challenges and opportunities for parental involvement in online mathematics learning for elementary school students.The New Educational Review69, 3 (20...

    2022

  38. [46]

    John TE Richardson. 2018. The use of Latin-square designs in educational and psychological research.Educational Research Review24 (2018), 84–97

    2018

  39. [47]

    B Rogoff. 1990. Apprenticeship in thinking: Cognitive development in social context

    1990

  40. [48]

    Vishnu Sarukkai, Lu Yuan, Mia Tang, Maneesh Agrawala, and Kayvon Fatahalian

  41. [49]

    InProceedings of the 37th Annual ACM Symposium on User Interface Software and Technology

    Block and detail: Scaffolding sketch-to-image generation. InProceedings of the 37th Annual ACM Symposium on User Interface Software and Technology. 1–13

  42. [50]

    Weiyan Shi, Hai Viet Le, and Kenny Tsu Wei Choo. 2025. Towards multimodal large-language models for parent-child interaction: A focus on joint attention. In Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–6

    2025

  43. [51]

    Yuling Sun, Jiaju Chen, Bingsheng Yao, Jiali Liu, Dakuo Wang, Xiaojuan Ma, Yuxuan Lu, Ying Xu, and Liang He. 2024. Exploring Parent’s Needs for Children- Centered AI to Support Preschoolers’ Interactive Storytelling and Reading Activ- ities.Proceedings of the ACM on Human-Comp...

    2024

  44. [52]

    H Lee Swanson, Olga Jerman, and Xinhua Zheng. 2008. Growth in working memory and mathematical problem solving in children at risk and not at risk for serious math difficulties.Journal of Educational Psychology100, 2 (2008), 343

    2008

  45. [53]

    Ronald D Taylor and Azeb Gebre. 2016. Teacher–student relationships and personalized learning: Implications of person and contextual variables.Handbook on personalized learning for states, districts, and schools(2016), 205–220

    2016

  46. [54]

    Maša Ðurišić and Mila Bunijevac. 2017. Parental involvement as a important factor for successful education.Center for Educational Policy Studies Journal7, 3 (2017), 137–153

    2017

  47. [55]

    Janneke Van de Pol, Monique Volman, and Jos Beishuizen. 2010. Scaffolding in teacher–student interaction: A decade of research.Educational psychology review 22, 3 (2010), 271–296

    2010

  48. [56]

    Devika Venugopalan, Ziwen Yan, Conrad Borchers, Jionghao Lin, and Vincent Aleven. 2025. Combining large language models with tutoring system intelli- gence: A case study in caregiver homework support. InProceedings of the 15th International Learning Analytics and Knowledge Con...

    2025

  49. [57]

    Lieven Verschaffel, Brian Greer, and Erik De Corte. 2000. Making sense of word problems. (2000)

    2000

  50. [58]

    Lieven Verschaffel, Stanislaw Schukajlow, Jon Star, and Wim Van Dooren. 2020. Word problems in mathematics education: A survey.Zdm52, 1 (2020), 1–16

    2020

  51. [59]

    1978.Mind in society: The development of higher psychological processes

    Lev Semenovich Vygotsky. 1978.Mind in society: The development of higher psychological processes. Vol. 86. Harvard university press

    1978

  52. [60]

    Isaac Wang, Pradyumna Narayana, Jesse Smith, Bruce Draper, Ross Beveridge, and Jaime Ruiz. 2018. Easel: Easy automatic segmentation event labeler. In Proceedings of the 23rd International Conference on Intelligent User Interfaces. 595–599

    2018

  53. [61]

    Wenjie Xu, Jiayi Ma, Jiayu Yao, Weijia Lin, Chao Zhang, Xuanhe Xia, Nan Zhuang, Shitong Weng, Xiaoqian Xie, Shuyue Feng, et al. 2023. Mathkingdom: Teaching children mathematical language through speaking at home via a voice-guided game. InProceedings of the 2023 CHI Conference...

    2023

  54. [62]

    Elinor’s talking to me!

    Ying Xu, Valery Vigil, Andres S Bustamante, and Mark Warschauer. 2022. “Elinor’s talking to me!”: Integrating conversational AI into children’s narrative science programming. InProceedings of the 2022 CHI conference on human factors in computing systems. 1–16

    2022

  55. [63]

    Riza Fatimah Zahrah and Winarti Dwi Febriani. 2020. A contextual problem based of local wisdom improve the ability to solving a word problem mathematics students of elementary school.PrimaryEdu: Journal of Primary Education4, 1 (2020), 55–64

    2020

  56. [64]

    Ali Zaidi and Karrie Karahalios. 2025. From Sociotechnical Gaps to Solutions: Designing AI Tools with Parents to Address Special Education Advocacy Barriers in IEP Processes. InProceedings of the 2025 ACM Designing Interactive Systems Conference. 2619–2636

    2025

  57. [65]

    Siyu Zha, Yuehan Qiao, Qingyu Hu, Zhongsheng Li, Jiangtao Gong, and Yingqing Xu. 2024. Designing child-centric AI learning environments: Insights from LLM- enhanced creative project-based learning.arXiv preprint arXiv:2403.16159(2024)

    2024 arXiv

  58. [66]

    Chao Zhang, Xuechen Liu, Katherine Ziska, Soobin Jeon, Chi-Lin Yu, and Ying Xu. 2024. Mathemyths: leveraging large language models to teach mathematical language through Child-AI co-creative storytelling. InProceedings of the 2024 CHI Conference on Human Factors in Computing S...

    2024

  59. [67]

    Jipeng Zhang, Lei Wang, Roy Ka-Wei Lee, Yi Bin, Yan Wang, Jie Shao, and Ee-Peng Lim. 2020. Graph-to-tree learning for solving math word problems. InProceedings of the 58th Annual Meeting of the Association for Computational Linguistics. 3928– 3937

    2020

  60. [68]

    Zheng Zhang, Ying Xu, Yanhao Wang, Bingsheng Yao, Daniel Ritchie, Tong- shuang Wu, Mo Yu, Dakuo Wang, and Toby Jia-Jun Li. 2022. Storybuddy: A human-ai collaborative chatbot for parent-child interactive storytelling with flex- ible parental involvement. InProceedings of the 20...

    2022

  61. [69]

    Yuhang Zhao, Alexis Hope, Jin Huang, Yoel Sumitro, James A Landay, and Yuanchun Shi. 2013. Hero: designing learning tools to increase parental in- volvement in elementary education in china. InCHI’13 Extended Abstracts on Human Factors in Computing Systems. 637–642. ParaTutor:...

    2013

  62. [70]

    Xinhua Zheng, H Lee Swanson, and George A Marcoulides. 2011. Working memory components as predictors of children’s mathematical word problem solving.Journal of experimental child psychology110, 4 (2011), 481–498. A Basic Information in the Formative Study Table 1: Basic Inform...

    2011

  63. [71]

    Effectiveness of Different Tutoring Modes • Did you observe any differences in your child’s understand- ing or performance across the different tutoring modes (A– D)? • Which mode seemed to support learning most effectively, and why? • Were there specific features (e.g., diagr...

  64. [72]

    User Experience with ParaTutor vs. DeepSeek • How did your experience differ between using the ParaTutor system (Modes C and D) and the DeepSeek system (Mode B)? • Which system provided clearer or more useful guidance to you as a parent? • Did you find ParaTutor’s visual and s...

  65. [73]

    Feasibility and Satisfaction with Tutoring Strategies • In ParaTutor, two modes were used: one aligning with your own strategies (Mode C), and one offering complementary strategies (Mode D). Which mode felt more feasible to im- plement in your home context? • Did the complemen...

  66. [74]

    Session 1 corresponds to Mode A and Test B

    General Suggestions and Reflections • What aspects of the system would you like to see improved (e.g., language clarity, pacing, visual design)? • Would you prefer AI to take a leading role in tutoring, or work alongside you as a co-tutor? • Would you recommend this system to ...

    2018

Pith tools

Reviewed July 15, 2026 · model on record in the stance chip above.