Dense retrieval models have predominantly been studied for English, where models have shown great success, due to the availability of human-labeled training pairs. However, there has been limited success for multilingual retrieval so far, as training data is uneven or scarcely available across multiple languages. Synthetic training data generation is promising (e.g., InPars or Promptagator), but has been investigated only for English. Therefore, to study model capabilities across both cross-lingual and monolingual retrieval tasks, we develop SWIM-IR, a synthetic retrieval training dataset containing 33 (high to very-low resource) languages for training multilingual dense retrieval models without requiring any human supervision. To construct SWIM-IR, we propose SAP (summarize-then-ask prompting), where the large language model (LLM) generates a textual summary prior to the query generation step. SAP assists the LLM in generating informative queries in the target language. Using SWIM-IR, we explore synthetic fine-tuning of multilingual dense retrieval models and evaluate them robustly on three retrieval benchmarks: XOR-Retrieve (cross-lingual), XTREME-UP (cross-lingual) and MIRACL (monolingual). Our models, called SWIM-X, are competitive with human-supervised dense retrieval models, e.g., mContriever, finding that SWIM-IR can cheaply substitute for expensive human-labeled retrieval training data.

翻译：稠密检索模型主要针对英语进行研究，由于拥有人工标注的训练数据对，这些模型在英语上取得了巨大成功。然而，多语言检索的研究进展有限，因为各语言之间的训练数据分布不均或极为稀缺。合成训练数据生成方法（如InPars或Promptagator）前景广阔，但迄今仅针对英语进行了探索。为此，为了研究模型在跨语言和单语言检索任务中的能力，我们开发了SWIM-IR——一个包含33种语言（从高资源到极低资源语言）的合成检索训练数据集，用于训练多语言稠密检索模型，且无需任何人工监督。为构建SWIM-IR，我们提出SAP（先摘要后提问提示）方法，其中大语言模型（LLM）在生成查询之前先生成文本摘要。SAP有助于LLM生成目标语言中信息丰富的查询。利用SWIM-IR，我们探索了多语言稠密检索模型的合成微调，并在三个检索基准上进行了稳健评估：XOR-Retrieve（跨语言）、XTREME-UP（跨语言）和MIRACL（单语言）。我们的模型SWIM-X与人工监督的稠密检索模型（如mContriever）相比具有竞争力，研究发现SWIM-IR可以低成本替代昂贵的人工标注检索训练数据。