Performance Gap in nDCG@10 for Code-Switched vs. Monolingual Models on BEIR Cross-Lingual Retrieval

Transferring information retrieval (IR) models from a high-resource language (typically English) to other languages in a zero-shot fashion has become a widely adopted approach. In this work, we show that the effectiveness of zero-shot rankers diminishes when queries and documents are present in different languages. Motivated by this, we propose to train ranking models on artificially code-switched data instead, which we generate by utilizing bilingual lexicons. To this end, we experiment with lexicons induced from (1) cross-lingual word embeddings and (2) parallel Wikipedia page titles. We use

Research goal: What is the performance gap (measured in nDCG@10) between models trained on artificially code-switched data and monolingual data when evaluated on the BEIR benchmark for cross-lingual retrieval?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.1/10.