Benchmark Belief Nets

These datasets were collected from the repository at: http://www.cs.huji.ac.il/labs/compbio/Repository/

These are all the BIF format ones from that site except for two: the Barley which requires specific premission for scientific use and the Diabetes one which I downloaded an incomplete version of.

The best source I could find for the BIF format is at http://www-2.cs.cmu.edu/~fgcozman/Research/InterchangeFormat/Old/xmlbif02.html. But the authors of the format seem to have moved on to some XML based thing so I'm keeping a local copy with UWML. I have extended the format in sevaral ways so as to parse the datasets from this repository.

Allow a string to begin with and contain _, -, %, or 0-9
Allow variable values to be numbers
Allow 0 and 1 to be probabilities (format required 0.0 and 1.0)

Some of these supposed BIF files do not comply with this version of the BIF format. UWML also doesn't deal with noisy functions, default values, or properties.

Belief Net Stats for alarm.bif:

Num Nodes: 37
Max Values: 4
Max Parents: 4
Max CPT Entries: 108

Belief Net Stats for insurance.bif:

Num Nodes: 27
Max Values: 5
Max Parents: 3
Max CPT Entries: 200

Belief Net Stats for Munin1.bif:

Num Nodes: 189
Max Values: 21
Max Parents: 3
Max CPT Entries: 600

Belief Net Stats for Munin2.bif:

Num Nodes: 1003
Max Values: 21
Max Parents: 3
Max CPT Entries: 600

Belief Net Stats for Munin3.bif:

Num Nodes: 1044
Max Values: 21
Max Parents: 3
Max CPT Entries: 600

Belief Net Stats for Munin4.bif:

Num Nodes: 1041
Max Values: 21
Max Parents: 3
Max CPT Entries: 600

Belief Net Stats for Water.bif:

Num Nodes: 32
Max Values: 4
Max Parents: 5
Max CPT Entries: 3072

Belief Net Stats for Link.bif:

Num Nodes: 724
Max Values: 4
Max Parents: 3
Max CPT Entries: 128

Belief Net Stats for Mildew.bif:

Num Nodes: 35
Max Values: 100
Max Parents: 3
Max CPT Entries: 280000

Belief Net Stats for hailfinder.bif:

Num Nodes: 56
Max Values: 11
Max Parents: 4
Max CPT Entries: 1188

Belief Net Stats for Pigs.bif:

Num Nodes: 441
Max Values: 3
Max Parents: 2
Max CPT Entries: 27

The following files can not be parsed by the current version of the BIF parser in UWML.

carpo.bif: Contains noisy functions