1Cademy - Multilingual Encoder Alone Does Not Improve MOOC-CS Recall (Hierarchical R@10 = 22.3 vs 23.1)

Learn Before

MOOC-CS: Language-Matched Controls (Results) in Auditable Strict-Parity Evaluation of Prerequisite-Graph Retrieval for RAG under Leakage Controls
Method Part 1: Multilingual Encoder + CJK Query Rewrite as a MOOC-CS Control (Auditable Strict-Parity Graph-RAG Paper)
Method Part 2: MOOC-CS Prerequisite Benchmark (Auditable Strict-Parity Graph-RAG Paper)

Example

Multilingual Encoder Alone Does Not Improve MOOC-CS Recall (Hierarchical R@10 = 22.3 vs 23.1)

On MOOC-CS ( $n = 103$ ) with the English-templated queries kept fixed, switching the dense encoder from English MiniLM to a multilingual model does not improve recall: hierarchical Recall@ $10$ moves from $23.1$ $[17.2, 29.3]$ to $22.3$ $[16.3, 29.0]$ , and Adaptive (heuristic) drops from $23.1$ to $19.9$ . The result isolates the effect of encoder choice alone under the unchanged English template interface and shows that a multilingual encoder, on its own, does not address the English-template / Chinese-concept-name mismatch.

0

1

Updated 2026-05-17

Contributors are:

Who are from:

References

Reference: Auditable Strict-Parity Evaluation of Prerequisite-Graph Retrieval for RAG under Leakage Controls

Learn Before

Related