Static Knowledge Grounding and Dynamic Personal Memory are complementary: removing both yields the largest quality degradation¶

Claim · All claims
Evidence · 1 study · causal r? · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? Ablation shows removing SKG mainly weakens grounding metrics while removing DPM mainly weakens Personalization and Fitness, and removing both yields the largest overall degradation. → Bingxi Zhao 2026

Evidence¶

Bingxi Zhao 2026¶

Bingxi Zhao, Jiahao Zhang, Xubin Ren, Zirui Guo, Tianzhe Chu, Yi Ma, Chao Huang. (2026). DeepTutor: Towards Agentic Personalized Tutoring. Technical Report, The University of Hong Kong. https://arxiv.org/abs/2604.26962 doi:10.48550/arxiv.2604.26962

q2 · i? · causal · r?

Component ablation (Figure 9) of SKG, DPM, or both from the full pipeline. The article reports SKG removal drops Groundedness most, while DPM removal most sharply declines Personalization and Fitness; no standardized effect sizes are printed.

"Removing DPMmainly weakens adaptation.PersonalizationandFitnessdecline most sharply, showing that the learner profile is central to gap-aware explanation and difficulty-calibrated practice; the smaller drop inGroundedness"

Discussion¶

Related Claims¶

• Ablating expert-informed structural components lowered LLM-judge scores on all dimensions, with largest drops in challenge quality, intent alignment, and domain grounding — related
• DeepTutor's interactive tutoring gains are stable across five university-level domains, varying only 0.16 points in overall quality — related
• Removing the cognitive load module causes the largest ablation performance drop, while state fusion has a smaller effect — related
• Pipeline components play complementary roles: hierarchical generation has the largest effect, grounding improves specificity and factual accuracy, and cluster-informed TOC planning improves structure and audience fit — related
• Knowledge and behavior profile components provide complementary signals: removing either hurts metrics tied to the other — related
• Removing explicit mastery information from learner profiles mainly degrades Adaptivity — related
• Each pedagogical verification component contributes to a distinct quality dimension: layout verification most strongly — related
• Ablation study shows metadata-enriched embeddings produce the largest accuracy drop, while attention masking, KV caching and dynamic batching primarily support inference efficiency — related
• Both CF-only and KG-only ablated variants underperform the full CF–KG model, indicating complementary behavioral and semantic signals — related
• Ablations show the trustworthiness module contributes most to performance, and the explanation generator cuts human review time by 41% — related
• DeepTutor improves overall first-person interactive tutoring quality by 10.76% over a Naive Tutor baseline on TutorBench — related