Multilingual Encoder + CJK Query Rewrite as a MOOC-CS Control
On MOOC-CS, the paper additionally runs a configuration that uses a multilingual encoder plus a CJK-only query rewrite to handle Chinese-character queries. This configuration is treated as a control experiment, not as a headline strict-parity comparator: because it changes the encoder (and therefore the parity contract), its numbers are reported separately and are used only to diagnose how much of the MOOC-CS gap is attributable to language alignment rather than to graph policy.
0
1
Tags
Science
Auditable Strict-Parity Evaluation of Prerequisite-Graph Retrieval for RAG under Leakage Controls
Related
Token-Cap Diagnostic Under a Shared Serialization Policy
Matched-Budget Best-First Traversal Comparator in Strict-Parity Prerequisite Retrieval
LectureBank-Full Paired Delta: Adaptive vs Hierarchical Baseline = +0.7 [-2.1, +3.6]
Multilingual Encoder + CJK Query Rewrite as a MOOC-CS Control
ColBERTv2/RePlug Reranking Baseline for Strict-Parity Prerequisite Retrieval
Truncated Personalized PageRank Baseline for Prerequisite Retrieval
Matched-Budget Best-First Traversal Comparator in Strict-Parity Prerequisite Retrieval
Multilingual Encoder + CJK Query Rewrite as a MOOC-CS Control
Multilingual Encoder + CJK Query Rewrite as a MOOC-CS Control
Learn After
Method Details for Multilingual Encoder + CJK Query Rewrite as a MOOC-CS Control (Auditable Strict-Parity Graph-RAG Paper)
Evaluation Evidence for Multilingual Encoder + CJK Query Rewrite as a MOOC-CS Control (Auditable Strict-Parity Graph-RAG Paper)
Scope Notes for Multilingual Encoder + CJK Query Rewrite as a MOOC-CS Control (Auditable Strict-Parity Graph-RAG Paper)