SYSTEMS AND METHODS FOR GENERATING LEADS IN A NETWORK BY PREDICTING PROPERTIES OF EXTERNAL NODES

• US 20110213741A1
• Filed: 08/10/2010
• Published: 09/01/2011
• Est. Priority Date: 08/10/2009
• Status: Active Grant

• Current Assignee: Mintigo Ltd.
• Original Assignee: Mintigo Ltd.
• Inventors: Ben-Reuven, Ehud, Sternfeld, Uri, Drukh, Evgeny, Segalov, Tal, Shama, Yaacov
• Granted Patent: US 8,560,471 B2
• Time in Patent Office: Days
• US Class Current: 706/13
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

The present invention is directed towards systems and methods for predicting one or more desired properties of external nodes or properties of their relations with internal nodes, based on a selected group of nodes about which it is known whether the nodes have the desired properties, or it is known whether they have a desired relation property with an internal node. The method comprises storing in one or more data structures a first data set regarding external nodes and a second data set regarding nodes with known properties in a selected group, each data set having one or more data items representing one or more events relating to or attributes of each node in the data set, the second data set including one or more types of data items not included in the first data set. The method then models the second data set to identify from the second data one or more modeled events or attributes of internal nodes in the selected group that are statistically likely to identify the nodes or their relations, that have the desired properties and predicts which of the external nodes are statistically likely to have the one or more desired properties, or desired relation property with internal node, based on the identified plurality of modeled events or attributes and the events or attributes in the first data set.

Claims

1. A computerized method for predicting one or more desired properties of external nodes based on a selected group of nodes about which it is known whether the nodes have the desired properties, the method comprising:

storing in one or more data structures a first data set regarding external nodes and a second data set regarding nodes in a selected group, each data set having one or more data items representing one or more events relating to or attributes of each node in the data set, the second data set including one or more types of data items not included in the first data set;

virtualizing the second data set regarding nodes into a modeled second data set after the first data set regarding external nodes at least by eliminating from the second data set the one or more data item types not included in the first data set;

modeling the virtualized second data set to identify from the modeled second data one or more modeled events or attributes of nodes in the selected group that are statistically likely to identify the nodes that have the desired properties; and

predicting which of the external nodes are statistically likely to have the one or more desired properties based on the identified plurality of modeled events or attributes and the events or attributes in the first data set.

2. The method of claim 1, wherein storing the second data set comprises storing data items regarding internal nodes that all are known to have the one or more desired properties.

3. The method of claim 1, wherein storing the second data set comprises storing data items regarding a positive set of nodes that are known to have the one or more desired properties and a negative set of nodes that are known to not have the one or more desired properties.

4. The method of claim 1, wherein storing the second data set comprises storing data items regarding a weighted set of nodes, the weighted set comprising a set of one or more desired properties.

5. The method of claim 3, wherein modeling the modeled second data set comprises identifying from the modeled second data set one or more first collections of events or attributes of nodes in the selected group that are statistically likely to make the nodes have the desired properties and one or more second collections of events or attributes of nodes in the selected group that are statistically likely to identify the nodes not have the desired properties.

6. The method of claim 5, wherein modeling the modeled second data set comprises identifying from the modeled second data set events or attributes that generate a statistically high distinction between the nodes in the positive set and the nodes in the negative set.

7. The method of claim 5, wherein statistically analyzing the modeled second data set comprises executing a machine learning algorithm program.

8. The method of claim 1, comprising generating the second data set from a larger data set.

9. The method of claim 8, wherein generating the second data set comprises selecting nodes from the larger data set based on one or more data items in the larger set having values identifying the corresponding nodes as having or not having the one or more desired properties.

10. The method of claim 1, wherein the internal nodes represent customers of a service provider and the external nodes represent entities that are not customers of the service provider, wherein the one or more desired properties comprise one or more properties of the non-customer entities that are specified by the service provider, and wherein storing the first and second data sets comprises storing data received from the service provider.

11. The method of claim 10, wherein storing the first and second data sets comprises storing transaction data regarding transactions involving customer and non-customer entities of the service provider.

12. The method of claim 11, wherein storing transaction data comprises storing transaction data for the second data set including one or more data items related to transactions involving the service provider customers, the data items only being receivable for transactions involving the service provider customers.

13. The method of claim 12, wherein the one or more data items related to transactions comprise transactions involving only service provider customers.

14. The method of claim 11, wherein the service provider comprises a telephone operator, and wherein storing transaction data comprises storing call detail records (CDRs) of telephone calls by or to customers and non-customer entities.

15. The method of claim 11, wherein the service provider comprises an e-mail provider, and wherein storing transaction data comprises storing e-mail records by or to users and non-users of the e-mail provider.

16. The method of claim 15, wherein storing e-mail records comprises storing one or more e-mail record data items selected from the group consisting of:

e-mail addresses, subject lines of e-mails, dates of e-mails, the number of e-mails sent or received, and a contact list associated with an e-mail.

17. The method of claim 16, wherein the one or more desired properties comprises properties of a given entity selected from the group consisting of:

volume of emails generated by the entity, friends of the entity that are non-users of the e-mail provider, social influence of a non-user of the email provider, and the entity having certain gender, age group, location, or occupation.

18. The method of claim 11, wherein the service provider comprises a social network operator, and wherein storing transaction data comprises storing records of relations or interaction between users and non-users of the social network provider.

19. The method of claim 18, wherein the one or more desired properties comprise properties of entities selected from the group consisting of:

social influence of a given user, number of connections a given user has, age of a given user, location of a given user, amount of time spent at the social network, gender of a given user, and the given user'"'"'s interest.

20. The method of claim 11, wherein the service provider comprises a delivery service provider, and wherein storing transaction data comprises storing delivery records between customers and non-customers of the delivery service provider.

21. The method of claim 20, wherein the one or more desired properties comprise properties of entities selected from the group consisting of:

customer names and addresses, customer billing information, and customer delivery orders.

22. The method of claim 14, wherein the one or more desired properties comprise properties of entities selected from the group consisting of:

entities being small offices/home offices, entities having high return per user, entities being customers adopting new technologies, and entities having certain gender, age group or location.

23. The method of claim 11, wherein the service provider comprises a financial institution, and wherein storing transaction data comprises storing logs of financial transactions between customers and non-customers.

24. The method of claim 23, wherein the one or more desired properties comprise properties of entities selected from the group consisting of:

account details, overall account value, and the number of transactions associated with an account.

25. The method of claim 11 wherein the entities within the selected group comprise a customers of a service provider.

26. The method of claim 25, comprising selecting service provider customers for the second data set based on service provider data identifying the selected customers as either having or not having the one or more desired properties.

27. The method of claim 25 comprising selecting service provider customers for the second data set based on service provider data identifying the selected customers as being connected to other nodes having one or more desired properties.

28. The method of claim 1, wherein virtualizing the second data set further comprises collecting transactions beginning or ending at the internal nodes to form a feature set for the nodes in the second set, but ignoring transactions that both begin and end at the second set.

29. The method of claim 28, wherein virtualizing the second data set further comprises normalizing the feature set based on an identified distribution of the external nodes.

30. The method of claim 29 wherein collecting transactions beginning or ending at the internal nodes to form a feature set for the internal nodes further comprises generating a feature profile for the internal nodes in the second set.

31. The method of claim 10, wherein virtualizing the second data set regarding nodes after the first data set regarding external nodes further comprises combining a plurality of nodes determined to be a member of a known subnet and ignoring all transactions within the known subnet.

32. The method of claim 31, wherein a known subnet comprises a plurality of nodes associated with a given family or business unit.

33. A system for predicting one or more desired properties of external nodes based on a selected group of internal nodes about which it is known whether the internal nodes have the desired properties, the system comprising:

a link pre-processor operative to analyze a first data set regarding external nodes and a second data set regarding internal nodes in a selected group;

one or more data structures storing the first data set regarding external nodes and the second data set regarding internal nodes in a selected group, each data set having one or more data items representing one or more events relating to or attributes of each node in the data set, the second data set including one or more types of data items not included in the first data set;

an intra-link virtualizer component operative to virtualize the second data set regarding internal nodes into a modeled second data set after the first data set regarding external nodes at least by eliminating from the second data set the one or more data item types not included in the first data set;

a learning machine operative to model the virtualized second data set to identify from the modeled second data set a plurality of modeled events or attributes of internal nodes in the selected group that are statistically likely to identify the internal nodes that have the desired properties;

a prediction module operative to predict the identified plurality of modeled events or attributes with the events or attributes in the first data set to predict which of the external nodes are statistically likely to have the one or more desired properties; and

a provider lead data storage unit operative to store the external nodes statistically likely to have the one or more desired properties.

34. The system of claim 33, wherein the link pre-processor is further operative to store data items regarding internal nodes that all are known to have the one or more desired properties in the one or more data structures.

35. The system of claim 33, wherein the link pre-processor is further operative to store data items regarding a positive set of internal nodes that are known to have the one or more desired properties and a negative set of internal nodes that are known to not have the one or more desired properties in the one or more data structures.

36. The system of claim 33, wherein the link pre-processor is further operative to store data items regarding a weighted set of internal nodes, the weighted set comprising a set of one or more desired properties in the one or more data structures.

37. The system of claim 36, wherein the learning machine is further operative to model the virtualized modeled second data set comprises identifying from the modeled second data set one or more first collections of events or attributes of internal nodes in the selected group that are statistically likely to identify the internal nodes that have the desired properties and one or more second collections of events or attributes of internal nodes in the selected group that are statistically likely to identify the internal nodes that don'"'"'t have the desired properties.

38. The system of claim 37, wherein the learning machine is further operative to identify from the modeled second data set events or attributes that generate a statistically high distinction between the internal nodes in the positive set and the internal nodes in the negative set.

39. The system of claim 33, wherein the link pre-processor is further operative to generate the second data set from a larger data set.

40. The system of claim 39, wherein the link pre-processor is further operative to select internal nodes from the larger data set based on one or more data items in the larger set having values identifying the corresponding internal nodes as having or not having the one or more desired properties.

41. The system of claim 33, wherein the internal nodes represent customers of a service provider and the external nodes represent entities that are not customers of the service provider, wherein the one or more desired properties comprise one or more properties of the non-customer entities that are more likely to make them become customers of the service provider, and wherein storing the first and second data sets comprises storing data received from the service provider.

42. The system of claim 41, wherein the one or more data structures further operative to store transaction data regarding transactions performed by the customer and non-customer entities of the service provider in the one or more data structures.

43. The system of claim 42, wherein the link pre-processor is further operative to store transaction data for the second data set including one or more data items related to transactions performed by the service provider customers, the data items only being receivable for transactions performed by the service provider customers in the one or more data structures.

44. The system of claim 43, wherein the one or more data items related to transactions comprise transactions between customers.

45. The system of claim 42, wherein the service provider comprises a telephone operator, and wherein the link pre-processor is further operative to store call detail records (CDRs) of telephone calls by or to customers and non-customer entities in the one or more data structures.

46. The system of claim 45, wherein the one or more desired properties comprise properties of entities selected from the group consisting of:

entities being family members, entities being small offices/home offices, entities having high return per user, and entities being customers adopting new technologies.

47. The system of claim 46, wherein the link pre-processor is further operative to select service provider customers for the second data set based on service provider data identifying the selected customers as either having or not having the one or more desired properties.

48. The system of claim 33 wherein the one or more desired properties includes a likelihood of acquisition of one of the external nodes.

49. The system of claim 33 wherein the intra-link virtualizer is further operative to collect transactions beginning or ending at the internal nodes to form a feature set for the internal nodes.

50. The system of claim 49 wherein the intra-link virtualizer is further operative to normalize the feature set based on an identified distribution of the external nodes.

51. The system of claim 49 wherein the intra-link virtualizer is further operative to generate a feature profile for the internal nodes.

52. The system of claim 42, wherein the service provider comprises an e-mail provider, and wherein the link pre-processor is further operative to store e-mail records by or to users and non-users of the e-mail provider.

53. The system of claim 42, wherein the service provider comprises a social network operator, and wherein the link pre-processor is further operative to store records of messages and relations between users and non-users of the social network provider.

54. The system of claim 42, wherein the service provider comprises a delivery service provider, and wherein the link pre-processor is further operative to store delivery customer profiles of customers and non-customers of the delivery service provider.

55. The system of claim 42, wherein the service provider comprises a financial institution, and wherein the link pre-processor is further operative to store customer account details and transaction logs.