Hugging Face's logo

Datasets:
Blaise-g
/
SumPubmed

Modalities:
Text
Formats:
parquet
Languages:
English
Size:
10K - 100K
Libraries:
Datasets
Dask
Dataset card Viewer Files Community
SumPubmed / README.md
Blaise-g's picture
Blaise-g
Update to enable 1 click evaluations
e1e1c9c
preview code
|
raw
history blame contribute delete
1.08 kB
metadata
language:
  - en
paperswithcode_id: null
pretty_name: SumPubmed
train-eval-index:
  - config: Blaise-g--SumPubmed
    task: summarization
    task_id: summarization
    splits:
      eval_split: test
    col_mapping:
      text: text
      abstract: target

Dataset Card for "SumPubmed"

Original Dataset Description

Description of dataset processing

5 rows were dropped from the original dataset taken from KAGGLE as they were missing the respective 'shorter_abstract' entries.

The 'line_text' and 'filename_text' columns were left untouched while the remaining ones were processed to remove the '\n' (many repetitions of those present in the original dataset), '<dig>', '<cit>', 'BACKGROUND', 'RESULTS' and 'CONCLUSIONS' matching strings which were deemed not necessary for the purpose of summarization. Additionally, extra spaces were removed and spacing around punctuations was fixed.