FUSE OGBN-MAG Embeddings: Academic Graph Node Vectors

OGBN-MAG is a heterogeneous academic graph from the Open Graph Benchmark. This dataset likely contains pre-computed node embeddings generated using the FUSE method after 200 training iterations. The embeddings were published on Kaggle, but the specific creation date and author are unknown.

Use Cases

• Benchmarking graph neural network models on the OGBN-MAG dataset (inferred from domain, verify after download)
• Performing node classification or author disambiguation tasks (inferred from domain, verify after download)
• Using pre-trained embeddings as features for downstream academic knowledge graph analysis (inferred from domain, verify after download)

Strengths

• Published on Kaggle, a platform with versioning and community discussion.
• Embeddings are specifically for the well-known OGBN-MAG benchmark graph.

Limitations

• Metadata is minimal; actual content requires verification after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and embedding dimensionality are unknown, which may limit suitability assessment.

Provenance

Source

Open Graph Benchmark (OGB) project.

Collection Method

Likely generated by applying the FUSE embedding algorithm to the OGBN-MAG graph.