Both CF-only and KG-only ablated variants underperform the full CF–KG model, indicating complementary behavioral and semantic signals¶

Claim · All claims
Evidence · 1 study · design r2 · 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? Removing the knowledge graph (CF-only, NDCG 0.584, HR 0.724) hurts more than removing collaborative filtering (KG-only, NDCG 0.604, HR 0.735); both fall below the full model's 0.625/0.751. → Bin Liu 2026

Evidence¶

Bin Liu 2026¶

Bin Liu, Xiaoli Sun, Jia Song. (2026). A hybrid collaborative filtering and knowledge graph-based cross-domain recommendation method for multimodal teaching resources. Computers and Education: Artificial Intelligence. https://doi.org/10.1016/j.caeai.2026.100678

q2 · i? · design · r2

Component ablation (Table 7) comparing CF-only, KG-only, and the full CF–KG model under the same dataset split and protocol. The article reports KG-only reaches NDCG 0.604 and HR 0.735, better than CF-only but below the full model; no standardized effect size is printed.

"The CF-only variant achieves an NDCG of 0.584 and an HR of 0.724, corresponding to decreases of 6.56% and 3.60%, respectively."

Discussion¶

Related Claims¶

• CF–KG achieves the best NDCG and hit rate on text, video, and interactive resource subsets, with modest but consistent gains over KGAT — related
• Removing any single modality degrades CF–KG only moderately, with text removal causing the largest drop (NDCG 0.625 to 0.593) — related
• Knowledge and behavior profile components provide complementary signals: removing either hurts metrics tied to the other — related
• Static Knowledge Grounding and Dynamic Personal Memory are complementary: removing both yields the largest quality degradation — related
• The CF–KG model attains the highest coverage (0.736) and cross-domain accuracy (0.741) among all compared recommenders — related
• The CF–KG hybrid model achieves the highest hit rate among all compared methods, rising from 0.718 to 0.751 across training — related
• The CF–KG hybrid model achieves higher offline NDCG than all eight baselines on the MARS English subset, reaching about 0.625 at epoch 40 — related