Dataset
big_patent
A corpus of around 1.3 million U.S. patent filings paired with human-authored abstractive summaries, organized into nine Cooperative Patent Classification categories ranging from human necessities to textiles and paper.
Data preview
Dataset card
Big Patent is a corpus of roughly 2.68 million U.S. patent records paired with human-written abstractive summaries, distributed under CC-BY-4.0. It is hosted on the Hugging Face hub by publisher manoj8890 and supports summarization research across nine Cooperative Patent Classification codes.
What this dataset contains
The dataset bundles U.S. patent applications, organized into nine Cooperative Patent Classification (CPC) code groupings such as 'a', 'b', and an 'all' umbrella configuration. Each row pairs the full Description section of a patent with its Abstract section, yielding a natural input-output setup for abstractive summarization. According to the metadata, the underlying pool covers 1.3 million records, with the released splits totaling about 2.68 million rows.
Structure and fields
Two string columns are exposed: 'description', extracted from the Description section of the patent application, and 'abstract', extracted from the Abstract section. Files ship in Parquet format, and the data is split into train, validation, and test partitions. The published card notes that older revisions of the dataset tokenize and lower-case these fields, while current defaults keep the cased raw strings.
How analysts use it
Users typically load the corpus via the Hugging Face datasets API to train and benchmark summarization models on long technical documents. Because each entry is tagged with a CPC code, the data can also be filtered to specific technology domains such as chemistry ('c') or textiles ('a') for domain-specific modeling. From an investment-research angle, the corpus offers a structured view of patented technology areas and could plausibly support thematic screens of innovation activity by classification, though any such use would require separate parsing of CPC codes into investable themes.
Documentation & references 1
Related datasets
Browse moreParsed eligibility criteria from 13,229 ClinicalTrials.gov studies representing the candidate pool gathered by two first-stage retrievers during TREC Clinical Trials 2021–2023, formatted as typed entity-relation graphs for reranking models.
The clinical-trials-patient-graphs dataset is a small tabular collection of patient-level clinical records distributed in Parquet format by the 2001jdev publisher. It organizes entities and relations extracted from trial data into structured rows spanning 2021 through 2023.
This dataset is a parsed snapshot of ClinicalTrials.gov records, distributed in Parquet format on the Hugging Face Hub under an unspecified license. It contains roughly 52,000 trial entries drawn from public registry filings.
clinical-trials-trec-qrels is a tabular relevance-judgment file distributed by the 2001jdev user on the Hugging Face Hub. It maps clinical-trial topic identifiers to NCT registry IDs with graded relevance scores.
The clinical-trials-trec-topics dataset is a small tabular collection of clinical-trial topic records distributed in Parquet format on the Hugging Face Hub. It appears to be derived from TREC clinical-trial retrieval topics, with rows partitioned by year.
SEC EDGAR 10-K & 10-Q financial statements for 2,600+ NASDAQ companies.
Discussion
Questions or observations about this dataset? Join the conversation on X.
No comments yet. Start the conversation on X.