Methods and apparatus for prediction and modification of behavior in networks

• US 20120303573A1
• Filed: 05/29/2012
• Published: 11/29/2012
• Est. Priority Date: 05/26/2011
• Status: Active Grant

• Current Assignee: Massachusetts Institute of Technology
• Original Assignee: Massachusetts Institute of Technology
• Inventors: Altshuler, Yaniv, Pentland, Alex Paul, Pan, Wei, Aharony, Nadav
• Granted Patent: US 9,098,798 B2
• Time in Patent Office: Days
• US Class Current: 706/52
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06Q 50/01 Social networking

Abstract

In exemplary implementations of this invention, mobile application (app) installations by users of one or more networks are predicted. Using network data gathered by smartphones, multiple “candidate” graphs (including a call log graph) are calculated. The “candidate” graphs are weighted by an optimization vector and then summed to calculate a composite graph. The composite graph is used to predict the conditional probabilities that the respective users will install an app, depending in part on whether the user'"'"'s neighbors have previously installed the app. Exogenous factors, such as the app'"'"'s quality, may be taken into account by creating a virtual candidate graph. The conditional probabilities may be used to select a subset of the users. Signals may be sent to the subset of users, including to recommend an app. Also, the probability of successful “trend ignition” may be predicted from network data.

Claims

1. A method of sending one or more signals, which method comprises, in combination:

(a) using one or more networks to gather network data, which one or more networks include node devices, and which network data includes data that is indicative of communications among at least some node devices, and

(b) using electronic components to send the one or more signals to a subset of human users of the one or more networks,

wherein

the subset comprises human users selected using an algorithm performed by one or more processors, and

the algorithm comprises

(i) using the network data to calculate a plurality of graphs,

(ii) using the plurality of graphs to calculate a composite graph,

(iii) using the composite graph to compute conditional probabilities, each of the probabilities, respectively, comprising a probability that a human user will engage in a specific behavior, and

(iv) using the conditional probabilities to select the subset of human users.

2. The method of claim 1, further comprising the step of using the one or more processors to perform the algorithm.

3. The method of claim 1, wherein a Type 1 Calculation is used to calculate the composite graph.

4. The method of claim 1, wherein a Type 5 Calculation is used to calculate the conditional probabilities.

5. The method of claim 1, wherein the conditional probability that the human user will engage in the specific behavior is calculated as being dependent, at least in part, on whether other human users that correspond to adjacent nodes in a graph have engaged in the same behavior.

6. The method of claim 1, further comprising using training data to calculate an optimal weight to assign to each graph, respectively, in the plurality of graphs.

7. The method of claim 1, further comprising using training data to solve an optimization problem by maximizing the conditional probabilities.

8. The method of claim 1, wherein a Type 6 Calculation is used for model training.

9. The method of claim 1, wherein the plurality of graphs includes a graph in which an edge between two nodes is weighted by a number of interactions between the two nodes.

10. The method of claim 9, wherein the interactions comprise phone calls.

11. The method of claim 9, wherein the interactions comprise messages that include text.

12. The method of claim 1, wherein the plurality of graphs includes a graph that corrects for an exogenous parameter of the behavior.

13. The method of claim 1, wherein the algorithm further comprises the step of calculating a trend ignition probability.

14. A method of sending one or more signals, which method comprises, in combination:

(a) using one or more networks to gather network data, which one or more networks include node devices, and which network data includes data that is indicative of communications among at least some node devices, and

(b) using electronic components to send the one or more signals to a subset of human users of the one or more networks,

wherein

the subset comprises human users selected using an algorithm performed by one or more processors, and

the algorithm comprises calculating a trend ignition probability.

15. The method of claim 14, further comprising the step of using the one or more processors to perform the algorithm.

16. The method of claim 14, wherein the algorithm includes a Type 11 Calculation.

17. The method of claim 14, wherein the algorithm includes a Type 12 Calculation.

18. A method comprising, in combination, the following steps:

(a) one or more computer processors;

(i) using network data regarding one or more networks to calculate a plurality of graphs,

(ii) using the plurality of graphs to calculate a composite graph,

(iii) using the composite graph to compute conditional probabilities, each of the probabilities, respectively, comprising a probability that a human user will engage in a specific behavior,

(iv) using the conditional probabilities to select a subset of human users of the one or more networks, and

(v) generating output data in machine-readable form, which output data identifies human users comprising or included in the subset, and

(b) one or more other electronic components sending the output data to (or for use in) a computer that is remote from the one or more computer processors.

19. The method of claim 18, wherein

the one or more other electronic components include a signal processor, and

the signal processor modifies a signal indicative of the output data.

20. The method of claim 18, wherein

the one or more other electronic components include a memory device, and

sending the output data comprises storing the output data in the memory device for delivery in any way to the computer that is remote.