ms2_sparse_mean

This is a copy of the [MS^2](https://huggingface.co/datasets/allenai/mslr2022) dataset, except the input source documents of its `validation` split have been replaced by a __sparse__ retriever. The retrieval pipeline used:

- __query__: The `background` field of each example
- __corpus__: The union of all documents in the `train`, `validation` and `test` splits. A document is the concatenation of the `title` and `abstract`.
- __retriever__: BM25 via [PyTerrier](https://pyterrier.readthedocs.io/en/latest/) with default settings
- __top-k strategy__: `"mean"`, i.e. the number of documents retrieved, `k`, is set as the mean number of documents seen across examples in this dataset, in this case `k==17`

Retrieval results on the `train` set:

| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.4333 | 0.2163 | 0.2051 | 0.2197 |

Retrieval results on the `validation` set:

| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.3780 | 0.1827 | 0.1815 | 0.1792 |

Retrieval results on the `test` set:

| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.3928 | 0.1898 | 0.1951 | 0.1820 |

本数据集为[MS^2](https://huggingface.co/datasets/allenai/mslr2022)的复刻版本，仅将其`validation`（验证集）划分的输入源文档替换为**稀疏检索器（sparse retriever）**的检索结果。所用检索流程如下：

- **查询（query）**：每个样本的`background`（背景）字段
- **语料库（corpus）**：`train`（训练集）、`validation`（验证集）与`test`（测试集）划分下的全部文档的并集。单篇文档由`title`（标题）与`abstract`（摘要）拼接而成。
- **检索器（retriever）**：基于[PyTerrier](https://pyterrier.readthedocs.io/en/latest/)实现的BM25检索，采用默认参数配置
- **top-k策略**：采用`"mean"`（均值）策略，即检索文档数`k`设置为当前数据集所有样本中平均出现的文档数量，本次任务中`k=17`。

训练集检索结果：
| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.4333 | 0.2163 | 0.2051 | 0.2197 |

验证集检索结果：
| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.3780 | 0.1827 | 0.1815 | 0.1792 |

测试集检索结果：
| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.3928 | 0.1898 | 0.1951 | 0.1820 |

搜集汇总

背景与挑战

背景概述

该数据集是MS^2数据集的衍生版本，其验证集的输入文档被稀疏检索器BM25替代，并采用均值top-k策略（k=17）进行检索。在训练、验证和测试集上，检索性能指标如Recall@100和Precision@k均被提供，例如验证集的Recall@100为0.3780。