# Are \`GNNHeteroGraph\` like labeled property graphs?

**URL:** <https://discourse.julialang.org/t/are-gnnheterograph-like-labeled-property-graphs/104131>\
**Category:** Graphs\
**Tags:** graphs, graphneuralnetworks\
**Created:** [September 22, 2023, 12:19am UTC](https://discourse.julialang.org/t/are-gnnheterograph-like-labeled-property-graphs/104131 "2023-09-22T00:19:34Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![huguesmp](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/huguesmp/32/202444_2.png) [@huguesmp](https://discourse.julialang.org/u/huguesmp)\
**Post date:** [September 22, 2023, 12:19am UTC](https://discourse.julialang.org/t/are-gnnheterograph-like-labeled-property-graphs/104131/1 "2023-09-22T00:19:34Z")

</div>

While reading [the documentation](https://carlolucibello.github.io/GraphNeuralNetworks.jl/dev/heterograph/) for `GraphNeuralNetworks.jl`, the section on heterogeneous graphs caught my eye.

With their typed nodes and relationships, they look and feel a little bit like labeled property graphs, as used in some graph databases such as [Neo4j](https://neo4j.com/fr/).

I was wondering if such a format/structure had been used outside of deep learning, and if so, if there are any examples or uses-cases for such graphs and ways to query and pattern match against them.