Contrastive Learning in Hyperbolic Space for Zero-Shot Cross-Lingual Retrieval Performance

This report synthesises findings from 6 peer-reviewed papers addressing the following research question: How does the performance of contrastive learning models in hyperbolic space for zero-shot cross-lingual retrieval vary with different language pairs in XOR-TyDi QA, measured by recall@k and NDCG. 8 claims were extracted from source literature; 8 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 8.5/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: How does the performance of contrastive learning models in hyperbolic space for zero-shot cross-lingual retrieval vary with different language pairs in XOR-TyDi QA, measured by recall@k and NDCG scores compared to Euclidean space?

Autonomous literature synthesis. Automated review score: 8.5/10. Full text and citation available at Assignee Research.