Useful experiments:

date

id

nodes

why?

2008-11-19 12:33:40

1227094419871

continuously growing output pool

1227198959811

self-adapting exit doors; RDFS inference Sesame; 3x unique triples produced WRONG: BUGGY HASH

2008-11-25 10:30:05

30

Fixed hash, Sesame reasoning, producing very little

2008-11-25 12:03:56

30

OWLIM reasoning, everything is stuck in the ?OutputPool, producing very little

2008-11-30 10:44:17

4 (virt)

Sesame reasoning, partitioned reading, read 1.5Mk, produced 2M

2008-11-30 10:58:48

4 (virt)

OWLIM reasoning, partitioned reading, read 1.5M, produced 1.55M

2008-12-05 08:33:37

40

OWLIM reasoning, partitioned reading wordnet, reasoner/input/output full and stable, dynamic exit-doors work well, but we're still slow producing (too many duplicates). We can see #iterations from triples_received / 100.000 (which they load in each iteration): 2.5 iterations/min

MarvinExperiments (last edited 2008-12-05 09:48:31 by EyalOren)