AlgorithmicResearchGroup/s2orc-cs-enriched

---
pretty_name: S2ORC CS Enriched
size_categories:
- 1M<n<10M
task_categories:
- text-classification
- summarization
language:
- en
tags:
- s2orc
- computer-science
- scientific-papers
- llm-enriched
---

# S2ORC CS Enriched

A Computer Science subset of the [Semantic Scholar Open Research Corpus (S2ORC)](https://github.com/allenai/s2orc) enriched with LLM-generated structured metadata. Contains 1.1 million CS papers with extracted methods, models, datasets, metrics, compute estimates, and summaries.

## Dataset Summary

| Statistic | Value |
|-----------|-------|
| Total papers | 1,117,706 |
| Total size | 54.7 GB |
| Parquet files | 1,118 |
| Split | `train` |

## Dataset Structure

### Base Columns

**Content**: `parsed_title`, `abstract`, `text`, `sections`, `figures`, `references`

**Authors & Dates**: `parsed_authors`, `metadata_authors`, `publicationdate`, `year`

**Identifiers**: `corpus_id`, `DOI`, `ArxIv`, `DBLP`, `ACL`, `CorpusId`, `PubMed`, `PubMedCentral`, `MAG`

**Publication Info**: `venue`, `journal`, `volume`, `pages`, `publicationtypes`, `publicationvenueid`

**Citations**: `citationcount`, `influentialcitationcount`, `referencecount`

**Open Access**: `isopenaccess`, `license`, `open_access_url`, `open_access_status`, `pdf_urls`, `pdf_sha`

**Classification**: `fieldofstudy_source`, `fieldofstudy_category` (all `Computer Science`)

### LLM Enrichment Columns

| Field | Description |
|-------|-------------|
| `summary` | Paper summary |
| `classification` | Paper classification (255 categories) |
| `methods` | Methods used |
| `results` | Key results |
| `models` | Models referenced |
| `datasets` | Datasets used |
| `metrics` | Evaluation metrics |
| `limitations` | Stated limitations |
| `explicit_gpu_hours` | GPU hours reported in the paper |
| `estimated_gpu_hours` | Estimated GPU hours |
| `number_gpu` | Number of GPUs used |
| `gpu_type` | GPU hardware type |
| `training_details` | Training configuration details |
| `reasoning` | Reasoning for compute estimates |

## Usage

```python
from datasets import load_dataset

# stream to avoid downloading 54.7GB
ds = load_dataset("AlgorithmicResearchGroup/s2orc-cs-enriched", streaming=True, split="train")
for paper in ds:
print(paper["parsed_title"])
print(f"Methods: {paper['methods']}")
print(f"GPU hours: {paper['estimated_gpu_hours']}")
break
```

## Related Resources

- [S2ORC ArXiv](https://huggingface.co/datasets/AlgorithmicResearchGroup/s2orc_arxiv) - Full ArXiv subset of S2ORC
- [Algorithmic Research Group - Open Source](https://algorithmicresearchgroup.com/opensource.html)

## Citation

```bibtex
@misc{s2orc_cs_enriched,
title={S2ORC CS Enriched},
author={Algorithmic Research Group},
year={2024},
publisher={Hugging Face},
url={https://huggingface.co/datasets/AlgorithmicResearchGroup/s2orc-cs-enriched}
}
```