Guide and Examples

This section contains a guide to the analytical engine and a number of examples.


We assume you has read the getting_started section and know how to launch a GraphScope session.

The analytical engine of GraphScope can handles various scenarios, … In most cases, the built-in algorithms are sufficient for your needs.

Using Louvain to detect communities in a social graph.

If you are a scientist familiar with NetworkX, you may find GraphScope works well for your known APIs to manipulate graph, to invoke an analysis algorithm, and it works well with other modules in PyData eco-system.

Using NetworkX to manipulate graph and run algorithms.

Running NetworkX Algorithms on GraphScope

If the built-in algorithms are not sufficient for your needs, you can implement your own algorithms in PIE or FLASH model, in Java、C++ or Python, and run them on GraphScope.

Write and run customized PageRank in C++ with PIE model

Write and run SSSP in C++ with FLASH model

Write and run algorithms in Python

Write and run algorithms in Java with PIE and Pregel model

Better still, if you already have your application running on Giraph or GraphX, the packaged jar can directly run on GraphScope. The migration is totally transparent, you even don’t need to have the source code!

Develop and run (existing) Giraph applications on GraphScope

Develop and run (existing) GraphX applications on GraphScope