[WIP] Implement bubbles (source and sink) pairs, for better clustering.#161

Open

Balletie wants to merge 23 commits intoProgrammingLife2015:masterfrom

Balletie:skip-cluster-bubble

Commits on Jun 19, 2015

First stab at creating bubbles, with source and sink labels and relationships
Balletie
committed
Replace test graph with one that conserves flow. Update tests accordingly
Balletie
committed
Remove visited set and keep track of processed rels using property
Balletie
committed
Create bubbles in PathExpander instead, correctly this time.
Balletie
committed
Test the bubbles with a new graph, update and enable old tests
Balletie
committed
Also test for multiple source nodes in bubbles. Currently fails
Balletie
committed
Add some comments, split some functions.

gfokkema
authored and
Balletie
committed
Add back propagateSourceIDs, and also store the propagatedSources for clustering
Balletie
committed
WIP: Initial clustering implementation for bubbles
Balletie
committed
Trim source and sink of clustered bubbles, return source and sink as singletons
Balletie
committed
Fix some perfomance issues: get rid of nested transactions, do explicit autoboxing
Balletie
committed
Test correctly, and fix bugs that arose (see description)
Balletie
committed
Set the interestingness property. Get the individual score in ClusterDrawable
Balletie
committed
Extend the test for individual nodes, and fix bug that arose
Balletie
committed
Performance improvement: use a query when clustering a large bubble
Balletie
committed

Commits on Jun 22, 2015

Continue clustering on visited nodes, pretend source is not there when outside of range
Balletie
committed