`vf2`

VF2 subgraph isomorphism algorithm in Rust

This crate implements the VF2 subgraph isomorphism algorithm [1]. It can find graph isomorphisms, subgraph isomorphisms, and induced subgraph isomorphisms. Graphs can be directed or undirected.

Features

[x] Enumerate graph isomorphisms
[x] Enumerate subgraph isomorphisms
[x] Enumerate induced subgraph isomorphisms
[x] Support directed graphs
[x] Support undirected graphs
[x] Support disconnected graphs
[x] Support node labels
[x] Support edge labels
[x] Graph trait

Motivation

Subgraph matching is the task of finding instances of a query graph within a larger data graph. It is useful when searching for patterns in a graph structure, such as relationships in a social network. It is a fundamental problem in graph theory with applications in pattern recognition, databases, security, and biochemistry.

Consider a network like LinkedIn where each node is a person, and each edge represents a connection. You are tasked with finding cases where five software developers and a doctor all know each other. You can make a query graph with developers and a doctor, and search for instances of that query in the larger network.

What is a subgraph isomorphism?

A graph is a structure consisting of a set of objects where some pairs of objects are connected. A graph isomorphism is a one-to-one correspondence between two graphs such that objects connected in one are also connected in the other.

Graph isomorphism

For two graphs to be isomorphic, there must be a one-to-one correspondence between nodes such that neighbors in one are also neighbors in the other. The query and data graphs in the following image are isomorphic.

Subgraph isomorphism

It is often desirable to find instances of one graph within another. To do this, we search for subgraph isomorphisms. A subgraph isomorphism is when one graph is isomorphic to a subgraph of another. There are two subgraph isomorphisms in the following image.

Induced subgraph isomorphism

An induced subgraph isomorphism is the same as a subgraph isomorphism except that the subgraph must be induced. Edges in the data subgraph must also exist in the query graph.

Usage

Add vf2 to your dependencies in Cargo.toml.

[dependencies]
vf2 = "1.0"

Create your query and data graphs with petgraph or any library that implements the Graph trait. Then, call one of the following functions based on the problem type.

| Problem type | Call | |-------------------------------|--------------------------------------| | Graph isomorphisms | vf2::isomorphisms | | Subgraph isomorphisms | vf2::subgraph_isomorphisms | | Induced subgraph isomorphisms | vf2::induced_subgraph_isomorphisms |

These return a Vf2Builder with the algorithm configured. Next, call one of the following on the builder to enumerate the isomorphisms.

| Desired output | Call | |--------------------------|---------| | First isomorphism | first | | Vector of isomorphisms | vec | | Iterator of isomorphisms | iter |

Filling a vector can consume a significant amount of memory. Use the iterator to inspect isomorphisms as they are found. For the best performance, call next_ref on the iterator instead of next to avoid cloning each isomorphism.

You can configure the node and edge equality functions on the builder with node_eq and edge_eq, respectively.

Example

This example shows how to find subgraph isomorphisms.

use petgraph::data::{Element, FromElements};
use petgraph::graph::DiGraph;

fn main() {
    // Create query graph.
    let query = DiGraph::<i32, ()>::from_elements([
        Element::Node { weight: 0 },
        Element::Node { weight: 1 },
        Element::Edge { source: 0, target: 1, weight: () },
    ]);

    // Create data graph.
    let data = DiGraph::<i32, ()>::from_elements([
        Element::Node { weight: 0 },
        Element::Node { weight: 1 },
        Element::Node { weight: 2 },
        Element::Edge { source: 0, target: 1, weight: () },
        Element::Edge { source: 1, target: 2, weight: () },
    ]);

    // Find subgraph isomorphisms.
    let isomorphisms = vf2::subgraph_isomorphisms(&query, &data).vec();
    assert_eq!(isomorphisms, vec![vec![0, 1], vec![1, 2]]);
}

Remaining work

The crate is feature complete. The following will improve performance.

[ ] Implement VF2 cutting rules
[ ] Implement VF2++ (only VF2 implemented so far)

References

[1] L. P. Cordella, P. Foggia, C. Sansone, and M. Vento, “A (sub)graph isomorphism algorithm for matching large graphs,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 26, no. 10, pp. 1367–1372, Oct. 2004, doi: https://doi.org/10.1109/tpami.2004.75.

[2] A. Jüttner and P. Madarasi, “VF2++—An improved subgraph isomorphism algorithm,” Discrete Applied Mathematics, vol. 242, pp. 69–81, Jun. 2018, doi: https://doi.org/10.1016/j.dam.2018.02.018.

Vf2

Install / Use

README