Multilingual Pre-trained Encoders Fine-tuned on Artificial Code-Switched Data for Zero-shot 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 impact of incorporating multilingual pre-trained language models (e.g., mBERT, XLM-R) as encoders in zero-shot cross-lingual retrieval systems when fine-tuned on artificially code-switched data, as measured by MRR and NDCG scores on the XGLUE benchmark?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 7.6/10.