The dataset viewer is not available for this split.
The size of the content of the first rows (465338 B) exceeds the maximum supported size (200000 B) even after truncation. Please report the issue.
Error code: TooBigContentError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Computer-science literature temporal graph
A typed temporal graph of the computer-science research literature (680,502 papers, 1950 to 2026-07-08; 7.29M dated edges; 5.07M resolved citations), together with the raw source dumps it was built from. Nodes are papers plus canonical methods, datasets, and tasks; edges are dated paper-to-entity mentions and citations, suitable for temporal link prediction and science-of-science studies.
Layout
raw/pwc/— Papers-with-Code archive, human-curated, frozen ~2025-06-30.raw/openalex/,raw/s2/,raw/s2_base/— a 2025–26 layer harvested from OpenAlex (arXiv ∩ CS) and enriched via the Semantic Scholar API (references, SPECTER v2 embeddings).processed/— the assembled typed graph in Parquet (nodes, typed edges, temporal splits undersplits/), plus precomputed entity profile embeddings (entity_emb_qwen*/).models/— small local checkpoints/artifacts used to build the graph.
Provenance and licensing
Derived from three upstream sources; downstream users must honour each:
- Papers-with-Code archive — CC-BY-SA 4.0 (this card inherits that licence for the base layer).
- OpenAlex — CC0.
- Semantic Scholar data (references, SPECTER v2) — subject to the Semantic Scholar API terms of use.
No personal data beyond public authorship metadata already present in the upstream sources.
- Downloads last month
- 60