---
license: cc-by-4.0
task_categories:
- graph-ml
language:
- en
pretty_name: Compact OpenAIRE Citation Graph
size_categories:
- 1B<n<10B
tags:
- citation-graph
- citation-network
- scholarly-data
- bibliometrics
- scientometrics
- openaire
- graph
- dynamic-graph
- temporal-graph
- dynamic-network
- link-prediction
configs:
- config_name: citations
  data_files: citations.parquet
- config_name: publications_large
  data_files: publications_large.parquet
---

[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.19207803.svg)](https://doi.org/10.5281/zenodo.19207803)

# Compact OpenAIRE Citation Graph

A distilled, compact representation of the complete OpenAIRE citation graph, making a
very large scholarly citation network easily accessible. The core graph is shared as two
files totalling ~11 GB; a larger node file with additional publication metadata is also
provided.

This is a working mirror for convenient loading. The **citable, archival record of
record is on Zenodo** ([10.5281/zenodo.19207803](https://doi.org/10.5281/zenodo.19207803)).
Please cite the accompanying data paper (see Citation below).

## Dataset structure

The dataset is a directed citation graph: publications are nodes, citations are edges.

| File | Role | Size (Parquet) |
| --- | --- | --- |
| `publications.parquet` | Nodes — publications with their primary DOI | 1.9 GB |
| `citations.parquet` | Edges — the citation links between publications | 8.8 GB |
| `publications_large.parquet` | Nodes with additional metadata fields (see below) | 72.5 GB |

### Fields in `publications_large`

| Field | Type | Description | Filled |
| --- | --- | --- | --- |
| `nodeId` | int32 | Unique internal node identifier | 100.00% |
| `openaireId` | str | OpenAIRE platform identifier | 100.00% |
| `title` | str | Publication title | 99.41% |
| `authors` | list[str] | Authors | 83.84% |
| `description` | str | Abstract / short description | 57.17% |
| `date` | datetime | Publication date | 97.33% |
| `container` | str | Journal / conference / repository | 68.45% |
| `citations` | int | Number of times cited | 97.33% |
| `language` | str | Language of the publication | 99.99% |
| `pid_dois` | list[str] | DOI identifiers | 80.70% |
| `pid_mag_ids` | list[str] | MAG IDs | 44.27% |
| `pid_pmids` | list[str] | PubMed IDs | 18.18% |
| `pid_handles` | list[str] | Persistent handles | 8.33% |
| `pid_pmcs` | list[str] | PubMed Central IDs | 4.77% |
| `pid_arxiv_ids` | list[str] | ArXiv IDs | 1.38% |

## Usage

The core node and edge files load comfortably in memory. Use the PyArrow backend:

```python
import pandas as pd

df_pubs = pd.read_parquet(
    "publications.parquet",
    engine="pyarrow",
    dtype_backend="pyarrow",
)

df_cites = pd.read_parquet(
    "citations.parquet",
    engine="pyarrow",
    dtype_backend="pyarrow",
)
```

`publications_large.parquet` will not fit in memory on most machines. Select only the
columns you need:

```python
df_large = pd.read_parquet(
    "publications_large.parquet",
    columns=["nodeId", "title", "pid_dois"],
    engine="pyarrow",
    dtype_backend="pyarrow",
)
```

You can also pull files directly from the Hub:

```python
from huggingface_hub import hf_hub_download

path = hf_hub_download(
    repo_id="USERNAME/compact-openaire-citation-graph",
    filename="citations.parquet",
    repo_type="dataset",
)
```

## Reproducibility

The PySpark pipeline used to produce these files (with a Singularity/Apptainer container
for portability) is archived on Zenodo as `pipeline.tar.xz` and maintained at
[Codeberg](https://codeberg.org/Zmeos/OpenAIRE-citation-extraction).

## License

Creative Commons Attribution 4.0 International (CC-BY-4.0).

## Citation

When using this dataset, please cite the accompanying article:

> Skarding, J. and Sanda, P. (2026) 'Making the Complete OpenAIRE Citation Graph Easily
> Accessible Through Compact Data Representation', *Journal of Open Humanities Data*,
> 12(1), p. 63. https://doi.org/10.5334/johd.520

```bibtex
@article{skarding2026openaire,
  author  = {Skarding, Joakim and Sanda, Pavel},
  title   = {Making the Complete OpenAIRE Citation Graph Easily Accessible Through Compact Data Representation},
  journal = {Journal of Open Humanities Data},
  volume  = {12},
  number  = {1},
  pages   = {63},
  year    = {2026},
  doi     = {10.5334/johd.520}
}
```

Dataset archive: Skarding, J. and Sanda, P. (2026). *Compact representation of the
OpenAIRE citation graph* (v2) [Data set]. Zenodo.
https://doi.org/10.5281/zenodo.19207803
---