Cross-lingual query decoding strategies and dense retrieval effectiveness on MLDoc

Effective cross-lingual dense retrieval methods that rely on multilingual pre-trained language models (PLMs) need to be trained to encompass both the relevance matching task and the cross-language alignment task. However, cross-lingual data for training is often scarcely available. In this paper, rather than using more cross-lingual data for training, we propose to use cross-lingual query generation to augment passage representations with queries in languages other than the original passage language. These augmented representations are used at inference time so that the representation can enco

Research goal: What is the impact of using different decoding strategies (e.g., beam search vs. top-k sampling) for cross-lingual query generation on the retrieval effectiveness of dense retrieval systems, as measured by nDCG@10 on the MLDoc benchmark?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.5/10.