Dataset
big_patent
An aggregation of 1.3 million U.S. patent records each accompanied by a human-written abstractive summary, sorted into nine Cooperative Patent Classification groupings spanning human necessities, chemistry, textiles, and related domains.
Data preview
Dataset card
big_patent is a public collection of roughly 1.3 million U.S. patent filings paired with human-written abstractive summaries, distributed in Parquet format under a CC-BY-4.0 license. It is designed primarily for training and evaluating summarization models over patent text.
What this dataset contains
The dataset covers U.S. patent applications organized under nine Cooperative Patent Classification (CPC) codes, available as a combined "all" split or individually per code. Each instance is a single patent with two string fields: `description`, drawn from the Description section of the filing, and `abstract`, drawn from the Abstract section. The current revision ships the 2.1.2 CPC taxonomy and preserves cased raw text rather than tokenized output.
Structure and fields
The released tables expose only two columns: `description` and `abstract`, both stored as string values. Data are provided across train, validation, and test splits totaling 2,682,724 rows in Parquet files. Code such as `load_dataset("big_patent", "a")` selects a single CPC bucket, while `concatenate_datasets` can combine splits across codes for multi-class training.
How analysts use it
For investment workflows, big_patent functions as a public-records source for monitoring innovation activity: the `description` and `abstract` fields can be mined to track filings by CPC code, detect emerging technology themes, or build text features tied to company-level patent portfolios. Its size makes it suitable for fine-tuning summarization models that condense long Description sections into the shorter Abstract field, supporting downstream alerting on new filings.
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.