Converting unordered graphs to oblivious read once ordered graph representation

• US 8,280,836 B2
• Filed: 09/08/2008
• Issued: 10/02/2012
• Est. Priority Date: 09/08/2008
• Status: Active Grant

• Current Assignee: Fair Isaac Corporation
• Original Assignee: Fair Isaac Corporation
• Inventors: Kumar, Prasun
• Primary Examiner(s): Gaffin, Jeffrey A
• Assistant Examiner(s): Hill, Stanley K
• Application Number:
  US12/206,514
• Publication Number:
  US 20100063953A1
• Time in Patent Office: 1,485 Days
• Field of Search: None
• US Class Current: 706/50
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation

Abstract

Data characterizing a desired variable order and an unordered graph can be received so that all paths in the unordered graph can be enumerated and a first path in the unordered graph can be leveled according to the desired variable order in a first oblivious read once decision graph. For each additional path other than the first path, the additional path is leveled in the desired order, nodes of the additional path are added to the first oblivious read once decision graph, and a union operation is performed on the first oblivious read once decision graph to union graph roots on the first oblivious read once decision graph. Thereafter, generation of a second oblivious read once decision graph can be initiated after completing processing of the first path and each additional path. Related apparatus, systems, techniques and articles are also described.

Claims

1. An article comprising a tangible machine-readable storage medium embodying instructions that when performed by one or more machines result in operations comprising:

receiving data characterizing a desired variable order and an unordered graph;

enumerating all paths in the unordered graph;

leveling, in an oblivious read once decision graph, a first path in the unordered graph according to the desired variable order; and

for each additional path in the unordered graph other than the first path, leveling the additional path in the desired order, adding nodes of the additional path to the oblivious read once decision graph, and performing a union operation on the oblivious read once decision graph to union graph roots on the oblivious read once decision graph.

2. An article as in claim 1, wherein each additional path added to the first oblivious read once decision graph is in a conjunctive normal form.

3. An article as in claim 1, wherein the first path is leveled in an empty oblivious read once decision graph.

4. An article as in claim 1, wherein the storage medium further embodies instructions that when performed by one or more machines result in operations comprising:

for each additional path, iterating each node in bottom up order to determine if same node already exists in the oblivious read once decision graph.

5. An article as in claim 4, wherein the storage medium further embodies instructions that when performed by one or more machines result in operations comprising:

adding a new node to the oblivious read once decision graph if is determined that such node does not already exist.

6. An article as in claim 4, wherein two nodes are considered same if corresponding node conditions and sub-graph below the two nodes are identical.

7. An article as in claim 6, wherein hashing techniques are used to identify isomorphic sub-graphs.

8. An article as in claim 1, wherein adding nodes for each additional path results in two root nodes in the oblivious read once decision graph, and wherein the union operation merges two root nodes and child nodes reachable from the root nodes such that there is only one root node in the oblivious read once decision graph and the child conditions of a parent are not overlapping.

9. An article as in claim 8, wherein the merging is performed such that any two nodes having the same subgraph are replaced with a new node with a condition that is union of the conditions of the replaced nodes.

10. An article as in claim 1, wherein the paths are enumerated by a depth first traversal.

11. A computer-implemented method comprising:

receiving data characterizing a desired variable order and an unordered graph;

enumerating all paths in the unordered graph;

leveling, in an oblivious read once decision graph, a first path in the unordered graph according to the desired variable order; and

for each additional path in the unordered graph other than the first path, leveling the additional path in the desired order, adding nodes of the additional path to the oblivious read once decision graph, and performing a union operation on the oblivious read once decision graph to union graph roots on the oblivious read once decision graph.

12. A method as in claim 11, wherein each additional path added to the first oblivious read once decision graph is in a conjunctive normal form.

13. A method as in claim 11, wherein the first path is leveled in an empty oblivious read once decision graph.

14. A method as in claim 11 further comprising:

for each additional path, iterating each node in bottom up order to determine if same node already exists in the oblivious read once decision graph.

15. A method as in claim 14 further comprising:

adding a new node to the oblivious read once decision graph if is determined that such node does not already exist.

16. A method as in claim 14, wherein two nodes are considered same if corresponding node conditions and sub-graph below the two nodes are identical.

17. A method as in claim 16, wherein hashing techniques are used to identify isomorphic sub-graphs.

18. A method as in claim 11, wherein adding nodes for each additional path results in two root nodes in the oblivious read once decision graph, and wherein the union operation merges two root nodes and child nodes reachable from the root nodes such that there is only one root node in the oblivious read once decision graph and the child conditions of a parent are not overlapping.

19. A method as in claim 18, wherein the merging is performed such that any two nodes having the same subgraph are replaced with a new node with a condition that is union of the conditions of the replaced nodes.

20. A method as in claim 11, wherein the paths are enumerated by a depth first traversal.

21. An article comprising a non-transitory machine-readable storage medium embodying instructions that when performed by one or more machines result in operations comprising:

receiving data characterizing a desired variable order and an unordered graph;

enumerating all paths in the unordered graph; and

recursively leveling, for an oblivious read once decision graph, each enumerated path according to the desired variable order.

22. An article as in claim 21, wherein the storage medium further embodies instructions that when performed by one or more machines result in operations comprising:

combining root nodes of the oblivious read once decision graph.