Extending English IR methods to multi-lingual IR

Abstract

This paper describes our participation in the 2023 WSDM CUP -MIRACL challenge. Via a combination of i) document translation; ii) multilingual SPLADE and Contriever; and iii) multilingual RankT5 and many other models, we were able to get first place in both the known and surprise languages tracks. Our strategy mostly revolved around getting the most diverse runs for the first stage and then throwing all possible reranking techniques. While this was not a first for many techniques, we had some things that we believe were never tried before, for example, we train the first SPLADE model that is effectively capable of working in more than 10 languages. However, a more careful study of the results is needed in order to verify if we were able to get first place just due to brute force or if the hybrids we developed really brought improvements over the other team's solutions.

Breaking Boundaries in Retrieval Systems: Unsupervised Domain Adaptation with Denoise-Finetuning

“…Two-stage retrieval (Matveeva et al, 2006;Liu et al, 2009;Wang et al, 2011;Yang et al, 2019) is a widely adopted approach that combines the strengths of retrieval models and rerank models for effective information retrieval. It has emerged as the preferred pipeline for competitive IR competition tasks (Lassance, 2023;Huang et al, 2023;Zhang et al, 2023).…”

mAggretriever: A Simple yet Effective Approach to Zero-Shot Multilingual Dense Retrieval

“…However, directly applying Eq. ( 2) to multilingual pre-trained MLM heads over the whole vocabulary increases computation cost for both training and inference due to large vocabulary size |V wp | in MLM projector, W mlm (Nair et al, 2022;Lassance, 2023). For example, the mBERT and XLM-R have respective vocabulary size of 120K and 250K (vs BERT's 35K) in the MLM projector.…”