---
license: cc0-1.0
size_categories:
- 1M<n<10M
configs:
- config_name: papers
  data_files: papers/**
- config_name: authors
  data_files: authors/**
task_categories:
- feature-extraction
- text-classification
- summarization
- text-generation
language:
- en
tags:
- nlp
- summarization
- networks
---

# Labeled-arXiv

Enriched arXiv paper metadata and derived author bibliometrics sourced from [OpenAlex](https://openalex.org/). Contains **2.93M papers** and **1.72M authors** across all arXiv subject areas.

## Subsets

### `papers` — 2.93M rows

| Column | Type | Description |
|---|---|---|
| `id` | `string` | arXiv paper ID (e.g. `0704.0028`) |
| `submitter` | `string` (nullable) | Name of the submitting author |
| `authors` | `string` | Raw author string from arXiv metadata |
| `title` | `string` | Paper title |
| `comments` | `string` (nullable) | Submitter comments (page count, figures, etc.) |
| `journal-ref` | `string` (nullable) | Journal publication reference |
| `doi` | `string` | Digital Object Identifier |
| `report-no` | `string` (nullable) | Report number |
| `categories` | `string` | arXiv category tags (e.g. `math.CA math.PR`) |
| `license` | `string` | License identifier (9 distinct values) |
| `abstract` | `string` | Paper abstract |
| `versions` | `list` | Version history entries |
| `update_date` | `date32` | Date of last metadata update |
| `authors_parsed` | `list` | Structured author names (parsed into components) |
| `author_ids` | `list` (nullable) | OpenAlex/ORCID author IDs linked to this paper |
| `deleted` | `bool` | Whether the paper has been removed from arXiv |
| `citations` | `list` (nullable) | List of citing work identifiers |
| `citation_count` | `int32` | Number of citations (range: 0–30.5k) |

### `authors` — 1.72M rows

Author-level metrics **derived from the papers in this dataset**, not global totals.

| Column | Type | Description |
|---|---|---|
| `author_id` | `string` | ORCID URL or OpenAlex ID |
| `name` | `string` (nullable) | Author display name |
| `paper_dois` | `list[string]` | DOIs of the author's papers in this dataset |
| `h_index` | `int32` | H-index computed within this dataset (0–84) |
| `works_count` | `int32` | Paper count within this dataset (1–13.9k) |
| `cited_by_count` | `int32` | Total citations within this dataset (0–57k) |

> Note: `h_index`, `works_count`, and `cited_by_count` are scoped to this dataset and do not represent an author's complete publication record.

## Usage

```python
from datasets import load_dataset

papers = load_dataset("ShayManor/Labeled-arXiv", "papers", split="train")
authors = load_dataset("ShayManor/Labeled-arXiv", "authors", split="train")
```

## Source

Metadata from [OpenAlex](https://openalex.org/) (Priem et al., 2022), built on top of the [arXiv](https://arxiv.org/) bulk metadata. Author IDs use [ORCID](https://orcid.org/) URLs or OpenAlex internal identifiers.

## License

[CC0 1.0](https://creativecommons.org/publicdomain/zero/1.0/) — Public Domain.