Heterogeneous Graph Benchmark - ACM

The ACM network as the Heterogeneous Graph Benchmark suite ships it, the same file as the acm entry. Contains papers, authors, subjects and terms, linked by citation, reference, authorship, subject and term relations, each listed in both directions. Used for benchmarking heterogeneous graph neural networks on node classification.

npx keepset fetch acm-hgb

Nodes: 10,942
Edges: 547,872
Size: 3.9 MiB

Fetch and preview this dataset on this page, or download the result as JSON. The command line fetches it without a browser.

Overview

Id	`acm-hgb`
Category	`graph/v1`
Kind	graph
Domain	Academic heterogeneous
Licence	not recorded by the source
Catalogue	`keepset` 1.33.0

Sources and provenance

#	URL	Status	Retrieved	Read by recipe	Pinned snapshot	SHA-256
0	https://drive.google.com/uc?export=download&id=1xbJ4QE9pcDJOcALv7dYhHDCPITX2Iddz	live	2026-03-26	yes

Citations

Lv et al. (2021). Are We Really Making Much Progress? A Worrying Analysis of Recent Neural Relation Extraction Models.

Recipe

Source	#0
Archive	zip
Part nodes	`ACM/node.dat`
Part links	`ACM/link.dat`
Parser format	`"hgb"`
Parser encoding	`"utf-8"`
Parser nodeTypes	`["paper","author","subject","term"]`
Parser linkTypes	`[{"name":"paper-cite-paper","source":"paper","target":"paper"},{"name":"paper-ref-paper","source":"paper","target":"paper"},{"name":"paper-author","source":"paper","target":"author"},{"name":"author-paper","source":"author","target":"paper"},{"name":"paper-subject","source":"paper","target":"subject"},{"name":"subject-paper","source":"subject","target":"paper"},{"name":"paper-term","source":"paper","target":"term"},{"name":"term-paper","source":"term","target":"paper"}]`

Crosswalk

keepset/typed to graph/v1, as a crossmap crosswalk.

Rule	From	To	Direction
move	`meta`	`meta`	both
move	`nodes[].id`	`nodes[].id`	both
move	`edges[].source`	`edges[].source`	both
move	`edges[].target`	`edges[].target`	both
move	`nodes[].type`	`nodes[].type`	both
move	`edges[].type`	`edges[].type`	both

Expected counts

Edges	547,872
Nodes	10,942

Properties

Components	1
Connected	yes
Cyclic	yes
Density	0.004576
Directed	yes
Download bytes	3.9 MiB
Edge types	8
Heterogeneous	yes
Multigraph	yes
Node types	4
Raw bytes	0 B
Self loops	yes
Weighted	no