caesar0301 / graphsim Goto Github PK

There are other repos that attemps something similar in https://github.com/hhchen1105/networkx_addon
Some other algorithm good to note:

• LHN https://github.com/hhchen1105/networkx_addon/blob/master/similarity/lhn.py
• RSS2 https://github.com/hhchen1105/networkx_addon/blob/master/similarity/rss2.py

I want to know that does GraphSim work for the problem as described below:

I have graph G1 and G2 and I want to check how much they are similar structure wise (as like isomorphism property of graph.)

Hi Chen,

thanks for GraphSim!

I used SimRank successfully on provided examples from the paper (university web graph, a bipartie graph of cake-bakers). However, when I try to run in on my graph which counts 144380 nodes and 223531 edges, I get the Memory Error Exception. The paper says they run the algorithm on 688,898 nodes 278,628 edges graph. Did you try your implementation on bigger graphs (real world graphs) than on provided examples?

Thanks,

I would like to see this package Python 3 compatible, so I made some changes. The most important change is the long->int in type decorators. At the moment this might reduce max iterations in Python 2 and I do not know serious problem is.

Examples now run with both Python 2 and Python 3. I hope this might help you.
Thanks

Hi, it would be awesome to have this package available as a conda package. Conda can distribute binaries, and that would ease the installation a lot!

I want to know how two graphs are structurally similar, instead of the content of edges and nodes. Irrespective of their node and edge similarity in terms of content. We required structural similarity.

Hi Chen,

Thank you for making graphsim. I was wondering if i could use it without root access on a machine.
Therefore, I compiled libtacsim locally and modified TACSim_in_C.py to find the local libs.

However, when i run print(gs.tacsim_combined_in_C(G1, G2)) my program crashes

I am using python2.7.14, graphsim is installed in a virtual environment venv, and here is the trace, any tips?

Traceback (most recent call last): File "compare_graph.py", line 22, in <module> print(gs.tacsim_combined_in_C(G1, G2)) File "/home/shoaloak/src/git/funky-nodes/venv/lib/python2.7/site-packages/graphsim/iter/TACSim_in_C.py", line 180, in tacsim_combined_in_C As, At = node_edge_adjacency(G1) File "/home/shoaloak/src/git/funky-nodes/venv/lib/python2.7/site-packages/typedecorator/__init__.py", line 389, in wrapper return fn(*args, **kwargs) File "/home/shoaloak/src/git/funky-nodes/venv/lib/python2.7/site-packages/graphsim/iter/TACSim.py", line 195, in node_edge_adjacency node_index[nodes[i]] = i TypeError: unhashable type: 'dict'

Greetings,

I've been trying to install graphsim on Ubuntu 16.04 but the main instructions don't seem to be for Ubuntu and I've tried several things but none work. I keep getting the error 'RuntimeError: Failed to build libtacsim'. I think I managed to install 'scons' but I am not sure since the reference command on this page isn't for Ubuntu...
Can I please get some directions on how to install graphsim on Ubuntu 16.04?

I used 'sudo apt-get install scons' and tried to install graphsim with 'sudo pip install -U graphsim' but I got that error.