How can I input my own graph data and compare it?

[Lost-little-dinosaur]

Hi I ran your train.py successfully, but I can't get an output and I want to use the script to compare my own graph data, how can I do? help me please~~.

[Lost-little-dinosaur]

and of course I gave you one star [doge]

[Lin-Yijie]

Hi~ You could construct your own dataset in train.py. The output of the model is exactly a graph representation as shown in 66 line of train.py.

[Lost-little-dinosaur]

I saw that I got the training set and the verification set in line 33, but it seemed to generate a seed, it just specified the limit range of the generated graph, and then generated an iterator through the GraphEditDistanceDataset class pairs function instead of input a concrete graph.

This is a little bit complicated for me, can you give me a concrete example where you input an adjacency matrix of two graphs, and you get an adjacency matrix of their similarity as an output? I would really appreciate it......

[Lin-Yijie]

Hi dinosaur~ In this repo, we give an example graph similarity learning task for synthesis graph data. It is highly expected to build your own graph dataset like GraphEditDistanceDataset. Similar to def _get_graph, you might use networkx package to change the adjacency matrix into the networkx like graph structure.

[Lin-Yijie]

@Lost-little-dinosaur Hi dinosaur. You could contact me with linyijie.gm@gmail.com

[eleveneee]

hi，bro Have you solved the problem of inputting your own graph? Also, how was your graph generated? Looking forward to your reply very much @Lost-little-dinosaur