transitiveclosure.txt

Efficient management of transitive relationships in large data and knowledge bases

predecessor / successor, immediate successor list

compression scheme: range compression
node numbering that yields maximum compression
	cover graph w/ spanning trees, use trees to generate node numbers

use interval indexes instead!
well, think that the algorithms below can be easily adapted for this actually

.. problem is that's not how they will be used in FRuleEngine
i.e., get all implied statements from cache and try them on rules


e.g., 
Activated , NotDone <-partOf- Active <-state:in- task1 , task2 , ..


(.. also need to support removals)

will this have such a large impact on performance in the end? ..

- transitive

map: predicate ->
	map: subjects (transitive closure)
	map: objects (co-occurring subjects)

predicates[predicate].subjects[subject] += object

if object in predicates[predicate].subjects
	predicates[predicate].subjects[subject] += predicates[predicate].subjects[object]

propagate-objects(subject)

fn propagate-objects(subject)
	subjects2 = predicates[predicate].objects[subject]
	for each subject2 in subjects2
		predicates[predicate].subjects[subject2] += predicates[predicate].subjects[subject]
		propagate-objects(subject2)


- property chain

map: predicate ->
	prior-predicate, next-predicate
	map: subjects (transitive closure)
	map: objects (co-occurring subjects)

predicates[predicate].subjects[subject] = object

if object in predicates[predicate.next-predicate].subjects
	predicates[predicate].subjects[subject] += predicates[predicate.next-predicate].subjects[object]

propagate-objects(predicate, subject)

fn propagate-objects(predicate, subject)
	subjects2 = predicates[predicate.prior-predicate].objects[subject]

	for each subject2 in subjects2
		predicates[predicate.prior-predicate].subjects[subject2] += predicates[predicate].subjects[subject]
		propagate-objects(predicate.prior-predicate, subject2)