PubMed 200k RCT: a Dataset for Sequential Sentence Classification in Medical Abstracts
This dataset addresses a bottleneck for researchers in medical fields who need tools to efficiently process long abstracts, though it is incremental as it builds on existing data sources.
The authors tackled the lack of large datasets for sequential sentence classification by creating PubMed 200k RCT, a dataset of 200,000 medical abstracts with 2.3 million sentences labeled for roles like background and result, aiming to improve algorithm accuracy and help researchers skim literature more efficiently.
We present PubMed 200k RCT, a new dataset based on PubMed for sequential sentence classification. The dataset consists of approximately 200,000 abstracts of randomized controlled trials, totaling 2.3 million sentences. Each sentence of each abstract is labeled with their role in the abstract using one of the following classes: background, objective, method, result, or conclusion. The purpose of releasing this dataset is twofold. First, the majority of datasets for sequential short-text classification (i.e., classification of short texts that appear in sequences) are small: we hope that releasing a new large dataset will help develop more accurate algorithms for this task. Second, from an application perspective, researchers need better tools to efficiently skim through the literature. Automatically classifying each sentence in an abstract would help researchers read abstracts more efficiently, especially in fields where abstracts may be long, such as the medical field.