Hi Andrew,
What I am looking for is the actually the deepest node. Without trying to give too much away. I am looking at patents and there is prior work on the topic though I think it did not fully exploit graph theory and my goal is strategies that use other ML techniques to create additional meta data that create new and additional relationships and additional strength to the graph. So with that premise.
I have two master nodes A & B and each of them will have certain classifications assigned to each of them with the number of classifications varying from 1 to as much as 200 though typically in the range of 2-10. The classifications are a pure tree graph. So for any two pair of master nodes what is the deepest deepest node for any pair of classification nodes. I actually don't need the whole path just the deepest node. The end point is a table of classification nodes for each pair and deepest node. I could impose a limit depth of 3 or from each side returning null if there is no common node within that number of levels. This will be input into other analysis tools.
Endpoint is a table for export.
Master A Master B Deepest
XXLB XXLA XXL
XXLBD XXLBF XXLB
XXLBD XXLA XXL
On Top of this is that the master nodes come from two holding groups and I would then do the pairwise look on each pair from the holding groups. There could be a fair amount of redundancy in the classification pairs so if there is a strategy of once the classification pair is identified with its deepest common node that result could be copied that may be a faster search strategy. But at the moment I working of looking if this search algorithm brings value.
Andy