cometadata/datacite-titles-descriptions-related-identifiers

---
license: cc0-1.0
task_categories:
- text-classification
- feature-extraction
language:
- multilingual
tags:
- datacite
- scholarly-metadata
- datasets
- research-data
size_categories:
- 10M<n<100M
---

# DataCite Dataset Titles, Descriptions, and Related Identifiers

Structured extraction of titles, descriptions, and related identifiers for all records with `resourceTypeGeneral: Dataset` in the DataCite metadata corpus.

## Source

Parsed from the **2026-03 DataCite Monthly Data File**.

## Contents

- **61,096,014 records** (one row per DOI)
- Filtered to `resourceTypeGeneral: Dataset` only
- Single Parquet file with nested columns

## Schema

| Column | Type | Description |
|--------|------|-------------|
| `doi` | `string` | The DOI identifier |
| `provider_id` | `string` | DataCite provider ID |
| `client_id` | `string` | DataCite client ID |
| `titles` | `list<struct>` | Array of `{title, titleType, lang}` |
| `descriptions` | `list<struct>` | Array of `{description, descriptionType, lang}` |
| `relatedIdentifiers` | `list<struct>` | Array of `{relatedIdentifier, relationType, relatedIdentifierType, resourceTypeGeneral}` |

## Usage

```python
import pyarrow.parquet as pq

table = pq.read_table("data/datasets_output.parquet")
print(table.schema)
print(f"{table.num_rows:,} records")
```

### Streaming with HuggingFace Datasets

```python
from datasets import load_dataset

ds = load_dataset("cometadata/datacite-titles-descriptions-related-identifiers", streaming=True)
for record in ds["train"]:
print(record["doi"], record["titles"])
break
```

## License

CC0 1.0 Universal - the metadata is from DataCite's open metadata corpus.