There is a class of graph distances, largely published about in the late 90s/early 2000s, based on maximum/minimum common subgraphs. Should we implement these?
The following papers claim that the computation of the maximum common subgraph is a special case of the computation of edit distance:
These papers are written by people more math-y than me. @leotrs (or anyone else!), could you take a look, specifically at 1 and 2, and comment on whether you think we should implement these?
I confess I haven't read the cited papers, but I have Things To Say.
We should leave open the possibility for implementing any distance method, so let's keep this open
I don't think we need to worry about this for the time being (let's revisit after NetSci'19)
One thing we could think about for the time being - do these distances fit the two-step model outlined in #174 ? If not, what changes would we need to make to the library to include these?
There is a class of graph distances, largely published about in the late 90s/early 2000s, based on maximum/minimum common subgraphs. Should we implement these?
The following papers claim that the computation of the maximum common subgraph is a special case of the computation of edit distance:
These define some variants of common subgraph based distances:
These papers are written by people more math-y than me. @leotrs (or anyone else!), could you take a look, specifically at 1 and 2, and comment on whether you think we should implement these?