liy140/multidomain-measextract-corpus

该数据集是一个多领域语料库，用于测量提取（Seq2Seq变体）。它包含了三个数据集（measeval、bm和msp）的训练、验证和测试数据。其中，measeval和msp数据集分别改编自MeasEval（Harper等人，2021）和Material Synthesis Procedual（Mysore等人，2019）语料库。该仓库将msp和measeval的提取任务聚合到段落级别，并以json格式提供标签，以便进行Seq2Seq训练。This dataset is a multi-domain corpus tailored for measurement extraction, a Seq2Seq variant task. It includes training, validation, and test splits for three datasets: measeval, bm, and msp. Specifically, the measeval and msp datasets are adapted from the MeasEval (Harper et al., 2021) and Material Synthesis Procedual (Mysore et al., 2019) corpora, respectively. This repository aggregates the extraction tasks of msp and measeval at the paragraph level, and provides labels in JSON format to facilitate Seq2Seq model training.

原始信息汇总

多领域测量提取语料库（Seq2Seq变体）

数据集配置

• measeval

  • 训练集: measeval_paragraph_level_no_spans_train.json
  • 验证集: measeval_paragraph_level_no_spans_val.json
  • 测试集: measeval_paragraph_level_no_spans_test.json

• bm

  • 训练集: bm_paragraph_level_no_spans_train.json
  • 验证集: bm_paragraph_level_no_spans_val.json
  • 测试集: bm_paragraph_level_no_spans_test.json

• msp

  • 训练集: msp_paragraph_level_no_spans_train.json
  • 验证集: msp_paragraph_level_no_spans_val.json
  • 测试集: msp_paragraph_level_no_spans_test.json

• all

  • 训练集:
    • measeval_paragraph_level_no_spans_train.json
    • bm_paragraph_level_no_spans_train.json
    • msp_paragraph_level_no_spans_train.json
  • 验证集:
    • measeval_paragraph_level_no_spans_val.json
    • bm_paragraph_level_no_spans_val.json
    • msp_paragraph_level_no_spans_val.json
  • 测试集:
    • measeval_paragraph_level_no_spans_test.json
    • bm_paragraph_level_no_spans_test.json
    • msp_paragraph_level_no_spans_test.json

任务类别

• 令牌分类

语言

• 英语

标签

• 化学
• 生物学

数据集大小

• n<1K