allenai/cochrane_sparse_mean

---
annotations_creators:
- expert-generated
language_creators:
- expert-generated
language:
- en
license:
- apache-2.0
multilinguality:
- monolingual
size_categories:
- 10K<n<100K
source_datasets:
- extended|other-MS^2
- extended|other-Cochrane
task_categories:
- summarization
- text2text-generation
paperswithcode_id: multi-document-summarization
pretty_name: MSLR Shared Task
---

This is a copy of the [Cochrane](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dataset, except the input source documents of its `validation` split have been replaced by a __sparse__ retriever.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 `target` 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==9`

Retrieval results on the `train` set:

| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.7014 | 0.3841 | 0.2976 | 0.4157 |

Retrieval results on the `validation` set:

| Recall@100 | Rprec | Precision@k | Recall@k |
| ----------- | ----------- | ----------- | ----------- |
| 0.7226 | 0.4023 | 0.3095 | 0.4443 |

Retrieval results on the `test` set:

N/A. Test set is blind so we do not have any queries.

原始信息汇总

数据集概述

基本信息

• 语言: 英语 (en)
• 许可证: Apache-2.0
• 多语言性: 单语种
• 大小: 10,000 < n < 100,000

数据来源与创建者

• 注释创建者: 专家生成
• 语言创建者: 专家生成
• 源数据集:
  • 扩展自 other-MS^2
  • 扩展自 other-Cochrane

任务类别

• 摘要生成 (summarization)
• 文本到文本生成 (text2text-generation)

数据集名称与标识

• 美观名称: MSLR Shared Task
• paperswithcode ID: multi-document-summarization

数据处理

• 验证集输入文档: 使用稀疏检索器替换
• 检索流程:
  • 查询: 每个示例的 target 字段
  • 语料库: train, validation, test 分割中所有文档的联合，文档由 title 和 abstract 串联而成
  • 检索器: 使用 PyTerrier 的 BM25，默认设置
  • top-k 策略: "mean"，即检索的文档数 k 设置为该数据集中示例平均看到的文档数，此处 k==9

检索结果

• 训练集:
  • Recall@100: 0.7014
  • Rprec: 0.3841
  • Precision@k: 0.2976
  • Recall@k: 0.4157
• 验证集:
  • Recall@100: 0.7226
  • Rprec: 0.4023
  • Precision@k: 0.3095
  • Recall@k: 0.4443
• 测试集: 无数据，因测试集为盲测，无查询信息