Mention counts below measure name matches in tracked papers, not benchmark quality or downloads. A family count can cover multiple datasets; it is not a separate count for each comparison row. Choose a task and version first, preserve evaluation splits, and follow the linked release terms. The site's content license does not replace the datasets' own terms.
Dataset comparison: tasks, access and usage conditions
Dataset / version
Task and evaluation
Access and preparation
Usage conditions
OGB — task-specific releases
Node, link and whole-graph property prediction across distinct datasets.
Evaluation: Use the task’s official metric and split (which may be temporal or scaffold-based); an OGB score is not one universal metric.
From Nov 2023 to Oct 2026, 86 archived papers mentioned Open Graph Benchmark in their title, abstract, keywords or full text, most of them in Aug 2025 (7).
No paper of the tracked journals mentioned Open Graph Benchmark in the last 30 days (2026-10-07 to 2026-11-05).
From 2017 to 2026 (so far), Open Graph Benchmark was mentioned in 92 papers in the tracked journals, most of them in AI (91) and information systems (1).
28 of them were published in 2025, against 0 in 2017.
Since 2023, the institutions with the most of these papers were Beijing University of Posts and Telecommunications (6), National University of Singapore (5) and Jingdong (China) (4).
The journals publishing most of them were IEEE Transactions on Pattern Analysis and Machine Intelligence (18), IEEE Transactions on Knowledge and Data Engineering (17) and IEEE Transactions on Neural Networks and Learning Systems (10).
The most-cited of them is “Large Language Models on Graphs: A Comprehensive Survey” (IEEE Transactions on Knowledge and Data Engineering, 2024), with 155 citations.
Papers use Open Graph Benchmark most often with Cora / CiteSeer / PubMed (Planetoid) (25), Molecule benchmarks (QM9, ZINC, MUTAG) (25) and Reddit (21); it has 18 partners in the code and data network, in the group around Twitter / X, Facebook, YouTube / TikTok.
Archive counts by publication year: a paper counts when its title, abstract or keywords name the dataset, or when OpenAlex's full-text search finds it there (searched for the last 30 days' papers and, by a pass over the past years, for older ones). Names added to the list later are looked for in older papers as the archive is read again.
Papers use Open Graph Benchmark most often with Cora / CiteSeer / PubMed (Planetoid) (25), Molecule benchmarks (QM9, ZINC, MUTAG) (25) and Reddit (21); it has 18 partners in the code and data network, in the group around Twitter / X, Facebook, YouTube / TikTok. The whole network →
What papers use Open Graph Benchmark together with (at least 3 papers; lift 1 = chance)
Scholar4AI. (2026). Open Graph Benchmark. Journal and Conference Research Rankings. Retrieved November 5, 2026, from https://scholar4ai.com/datasets/open-graph-benchmark
@misc{scholar4ai-datasets-open-graph-benchmark,
author = {{Scholar4AI}},
title = {Open Graph Benchmark},
year = {2026},
howpublished = {\url{https://scholar4ai.com/datasets/open-graph-benchmark}},
urldate = {2026-11-05},
note = {Journal and Conference Research Rankings; data from OpenAlex (CC0); licensed CC BY-NC 4.0}
}
Free to reuse for non-commercial purposes with attribution (CC BY-NC 4.0).