Random Walks (Weisfeiler-Lehman)

This library provides an implementation of the Weisfeiler-Lehman Graph Kernel based on graph2vec.

Interface

To use this library, prefix the types and functions below with WlKernel..

Methods

• val wlGraph: ((int * int) * (int list)) list list * int -> word list list

Method Overview

• wlGraph (graph, depth)
  • Random walk to embed graphs. The first argument is a collection of graphs, each graph is represented a list, each line in list includes the index of node in the graph, the information bound, and a list of int represents its neighbors’ indices. The second argument is the depth of walking. The result is a list of embedding of input graphs. Each graph is embedded as a list of word (See graph2vec for details).