Dependency graph in data-driven model

• US 8,352,397 B2
• Filed: 09/10/2009
• Issued: 01/08/2013
• Est. Priority Date: 09/10/2009
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Rubin, Darryl E., Mital, Vijay, Beckman, Brian C., Katzenberger, G. Shon
• Primary Examiner(s): VINCENT, DAVID ROBERT
• Application Number:
  US12/557,297
• Publication Number:
  US 20110060704A1
• Time in Patent Office: 1,216 Days
• Field of Search: 706/12, 706/45, 706/62
• US Class Current: 706/45
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/04 Inference or reasoning models

Abstract

The inference of a dependency graph that represents a graph of solves that leads from input model parameter(s) to output model parameters using analytics. In one embodiment, the dependency graph is part of visually driven analytics in which the output model parameter(s) are used to formulate data-drive scenes. As the identity of the input and/or output model parameter(s) change, or as the analytics themselves change, the dependency graph may also change. This might trigger a resolve of the analytics. In one embodiment, the intermediate parameters involved in the dependency graph may be viewed and evaluated by the user.

Claims

1. A physical computer program product comprising one or more physical computer-readable media having thereon computer-executable instructions that are structured such that, when executed by one or more processors of a computing system, the computing system is caused to infer a dependency graph by performing the following:

an act of identifying one or more input model parameter(s) for a particular solve;

an act of identifying one or more output model parameter(s) for the particular solve, wherein the identity of the input model parameter(s) or the output model parameter(s) may change from one solve to the next;

an act of identifying a first model analytical expression that will allow a solve for at least one of the output model parameter(s) using a first set of one or more other parameters, the first model analytical expression selected from a pool of model analytical expressions corresponding to an analytical model;

an act of formulating at least a part of a dependency graph that identifies that the first set of one or more parameters in conjunction with the identified first model analytical expression leads to the at least one of the output model parameter(s), wherein at least one of the first set of one or more parameters is an intermediate model parameter, and not one of the input model parameter(s);

an act of identifying a second model analytical expressions that will allow for a solve for the intermediate model parameter given a second set of one or more parameters, and an act of extending the dependency graph to reflect that the second set of one or more parameters in conjunction with the identified second model analytical expression leads to the intermediate model parameter; and

if there is any in the second set of one or more parameters that is an intermediate model parameter, an act of recursively repeating the act of identifying the analytics and extending the dependency graph until the dependency graph shows a solve dependency that leads from the identified one or more input model parameter(s) to the identified one or more output model parameter(s) through a plurality of intermediate model parameters for the particular solve.

2. The physical computer program product in accordance with claim 1, wherein the dependency graph includes a plurality of nodes representing parameters, and includes a plurality of links representing analytical relations between the parameters.

3. The physical computer program product in accordance with claim 2, wherein the computer-executable instructions are structured such that the computing system provides a user interface in which the user can view properties associated with at least some nodes in the dependency graph.

4. The physical computer program product in accordance with claim 3, wherein for at least one of the at least some nodes in the dependency graph, the properties include a value of a parameter associated with the node.

5. The physical computer program product in accordance with claim 4, wherein for the at least one of the at least some nodes in the dependency graph, the properties also include an identity of a parameter associated with the node.

6. The physical computer program product in accordance with claim 3, wherein for at least one of the at least some nodes in the dependency graph, the properties include an identity of a parameter associated with the node.

7. The physical computer program product in accordance with claim 6, wherein the computer-executable instructions are structured such that the computing system provides a user interface in which the user can view properties associated with at least some links in the dependency graph.

8. The physical computer program product in accordance with claim 7, wherein at least one property associated with the at least some of the links includes an analytical expression.

9. The physical computer program product in accordance with claim 2, wherein the computer-executable instructions are structured such that the computing system provides a user interface in which the user can view properties associated with at least some links in the dependency graph.

10. The physical computer program product in accordance with claim 9, wherein at least one property associated with the at least some of the links includes an analytical expression.

11. The physical computer program product in accordance with claim 2, wherein the computer-executable instructions are structured such that the computing system provides a user interface in which the user can edit properties associated with at least some nodes in the dependency graph.

12. The physical computer program product in accordance with claim 11, wherein the computer-executable instructions are structured such that the computing system provides a user interface in which the user can edit properties associated with at least some links in the dependency graph.

13. The physical computer program product in accordance with claim 2, wherein the computer-executable instructions are structured such that the computing system provides a user interface in which the user can edit properties associated with at least some links in the dependency graph.

14. The physical computer program product in accordance with claim 2, wherein the computer-executable instructions are structured such that the computing system provides a user interface in which the user can edit a structure of the dependency graph.

15. A method for inferring a dependency graph, the method comprising the following:

an act of a physical computing system identifying one or more input model parameter(s) for a particular solve to be performed by the physical computing system;

an act of the physical computing system identifying one or more output model parameter(s) for the particular solve, wherein the identity of the input model parameter(s) or the output model parameter(s) may change from one solve to the next;

an act of the physical computing system identifying a first model analytical expression that will allow a solve for at least one of the output model parameter(s) using a first set of one or more other parameters, the first model analytical expression selected from a pool of model analytical expressions corresponding to an analytical model;

an act of the physical computing system formulating at least a part of a dependency graph that identifies that the first set of one or more parameters in conjunction with the identified first model analytical expression leads to the at least one of the output model parameter(s), wherein at least one of the first set of one or more parameters is an intermediate model parameter, and not one of the input model parameter(s);

an act of the physical computing system identifying a second model analytical expressions that will allow for a solve for the intermediate model parameter given a second set of one or more parameters, and an act of extending the dependency graph to reflect that the second set of one or more parameters in conjunction with the identified second model analytical expression leads to the intermediate model parameter; and

if there is any in the second set of one or more parameters that is an intermediate model parameter, an act of the computing system recursively repeating the act of identifying the analytics and extending the dependency graph until the dependency graph shows a solve dependency that leads from the identified one or more input model parameter(s) to the identified one or more output model parameter(s) through a plurality of intermediate model parameters for the particular solve.

16. The method in accordance with claim 15, wherein the dependency graph includes a plurality of nodes representing parameters, and includes a plurality of links representing analytical relations between the parameters.

17. The method in accordance with claim 16, wherein the computing system provides a user interface in which the user can view or edit properties associated with at least some nodes in the dependency graph.

18. The method in accordance with claim 17, wherein the computing system provides a user interface in which the user can view or edit properties associated with at least some links in the dependency graph.

19. The method in accordance with claim 16, wherein the computing system provides a user interface in which the user can view or edit properties associated with at least some links in the dependency graph.

20. A physical computer program product comprising one or more physical computer-readable media having thereon computer-executable instructions that are structured such that, when executed by one or more processors of a computing system, the computing system is caused to infer a dependency graph by performing the following:

an act of identifying one or more input model parameter(s) for a particular solve;

an act of identifying one or more output model parameter(s) for the particular solve, wherein the identity of the input model parameter(s) or the output model parameter(s) may change from one solve to the next;

an act of identifying a first model analytical expression that will allow a solve for at least one of the output model parameter(s) using a first set of one or more other parameters, the first model analytical expression selected from a pool of model analytical expressions corresponding to an analytical model;

an act of formulating at least a part of a dependency graph that identifies that the first set of one or more parameters in conjunction with the identified first model analytical expression leads to the at least one of the output model parameter(s), wherein at least one of the first set of one or more parameters is an intermediate model parameter, and not one of the input model parameter(s);

an act of identifying a second model analytical expressions that will allow for a solve for the intermediate model parameter given a second set of one or more parameters, and an act of extending the dependency graph to reflect that the second set of one or more parameters in conjunction with the identified second model analytical expression leads to the intermediate model parameter; and

if there is any in the second set of one or more parameters that is an intermediate model parameter, an act of recursively repeating the act of identifying the analytics and extending the dependency graph until the dependency graph shows a solve dependency that leads from the identified one or more input model parameter(s) to the identified one or more output model parameter(s) through a plurality of intermediate model parameters for the particular solve,

wherein the dependency graph includes a plurality of nodes representing parameters, and includes a plurality of links representing analytical relations between the parameters,

wherein the computer-executable instructions are structured such that the computing system provides a user interface in which the user can view or edit properties associated with at least some nodes in the dependency graph, and can view or edit properties associated with at least some links in the dependency graph.