Systems and methods for inferring entity relationships via network communications of users or user devices
A number of techniques facilitate generation of data points from observations about network traffic. An inferencing system can use these data points to determine whether a relationship exists between two entities or whether an existing relationship has terminated, without any external knowledge of the existence of or termination of such a relationship.
1. A method for inferring a relationship between two entities, the method comprising the steps of:
receiving at a server a network observation dataset, each data point in the dataset comprising (a) a device identifier, and (b) (i) a network address or (b) (ii) a device location;
selecting from the dataset a first-type data point comprising: (A) a first device identifier, and (B)(i) a first network address or (B)(ii) a first device location;
identifying a first entity associated with the first network address or the first device location;
selecting from the dataset a second-type data point comprising: (A) the first device identifier, and (C)(i) a second network address different from the first network address or (C)(ii) a second device location different from the first device location;
identifying a second entity that is different from the first entity and that is associated with the second network address or the second device location;
inferring, at a baseline confidence level, a relationship between the first entity and the second entity;
determining that: a total number or frequency of the first-type data points in the dataset is at least equal to a specified home-count or home-frequency threshold; and
determining that: a total number or frequency of the second-type data points in the dataset is at least equal to a specified visitor-count or visitor-frequency threshold.
2. The method of claim 1 , wherein: the first network address comprises: a first Internet Protocol (IP) address, a first service set identifier (SSID), or a first basic SSID (BS SID); and the second network address comprises a second IP address, a second SSID, or a second BSSID.
3. The method of claim 1 , further comprising:
selecting from the dataset a third-type data point comprising: (D) a second device identifier, and (C)(i) the second network address or (C)(ii) the second device location;
selecting from the dataset a fourth-type data point comprising: (D) the second device identifier, and (B)(i) the first network address or (B)(ii) the first device location; and increasing the baseline confidence level.
4. The method of claim 1 , further comprising:
increasing the baseline confidence level.
5. The method of claim 1 , further comprising:
receiving an enrichment attribute; testing whether the enrichment attribute is associated with the first entity or the second entity; and
increasing the baseline confidence level, if the enrichment attribute is determined to be associated with the first entity or the second entity; and
otherwise, decreasing the baseline confidence level or invalidating the inferred relationship.
6. The method of claim 5 , wherein:
the enrichment attribute comprises an entity identifier; and the testing step comprises determining whether the entity identifier identifies the first entity, the second entity, or a different third entity.
7. The method of claim 5 , wherein:
the enrichment attribute comprises a service set identifier (SSID); and
testing step comprises determining whether the SSID is associated with the first network address or the second network address or neither the first nor the second network address.
8. The method of claim 5 , wherein the enrichment attribute comprises a physical location associated with the first or the second network address.
9. The method of claim 5 , wherein:
the enrichment attribute comprises a classification of the first or the second network address; and
the testing step comprises determining whether the classification indicates a business entity.
10. The method of claim 5 , wherein:
the enrichment attribute comprises a list of Access Points (APs) associated with the first network address; and the testing step comprises determining whether one or more APs in the list are designated as non-guest APs.
11. The method of claim 5 , wherein:
the enrichment attribute comprises a list of Access Points (APs) associated with the second network address; and the testing step comprises determining whether one or more APs in the list are designated as guest APs.
12. The method of claim 5 , wherein:
the enrichment attribute comprises a list of devices associated with the first or the second network address and a respective type of each device; and the testing step comprises determining whether one or more devices in the list are endpoint devices.
13. The method of claim 5 , wherein the first data point or the second data point comprises the enrichment attribute.
14. A system for inferring a relationship between two entities, comprising:
a processor;
a network port in communication with the processor and adapted to receive from a data source a network observation dataset, each data point in the dataset comprising one or more of: (i) a device identifier, (ii) a user identifier, (iii) a network address, or (iii) a device location; and
a memory coupled to the processor and comprising instructions, which when executed by the processor, program the processor to:
select from the dataset a first-type data point comprising: (A)(i) a first device identifier or (A)(ii) a first user identifier, and (B)(i) a first network address or (B)(ii) a first device location;
identify a first entity associated with the first network address or the first device location;
select from the dataset a second-type data point comprising: (A)(i) the first device identifier or (A)(ii) the first user identifier, and (C)(i) a second network address or (C)(ii) a second device location;
identify a second entity associated with the second network address or the second device location;
infer a relationship between the first entity and the second entity and designate to the inferred relationship a baseline confidence level;
determining that: a total number or frequency of the first-type data points in the dataset is at least equal to a specified home-count or home-frequency threshold; and
determining that: a total number or frequency of the second-type data points in the dataset is at least equal to a specified visitor-count or visitor-frequency threshold.
15. The system of claim 14 , wherein:
the network port is adapted to receive an enrichment attribute; and the instructions further program the processor to: test whether the enrichment attribute is associated with the first entity or the second entity; and increase the baseline confidence level, if the enrichment attribute is determined to be associated with the first entity or the second entity; and otherwise, decrease the baseline confidence level or invalidate the inferred relationship.
16. The system of claim 15 , wherein the enrichment attribute comprises: (i) an entity identifier, (ii) a service set identifier (SSID), (iii) a physical location associated with the first or the second network address, (iv) a classification of the first or the second network address, (v) a list of Access Points (APs) associated with the first or the second network address and a respective type of each AP, or (vi) a list of devices associated with the first or the second network address and a respective type of each device.
17. The system of claim 15 , wherein the first or the second data point comprises the enrichment attribute.
18. A method for inferring a relationship between two entities, the method comprising the steps of:
receiving at a server a network observation dataset, each data point in the dataset comprising a user identifier and a network address;
selecting from the dataset a first-type data point comprising a first user identifier and a first network address;
identifying a first entity associated with the first network address; selecting from the dataset a second-type data point comprising the first user identifier and a second network address different from the first network address; identifying a second entity that is different from the first entity and that is associated with the second network address;
inferring at a baseline confidence level a relationship between the first entity and the second entity;
determining that: a total number or frequency of the first-type data points in the dataset is at least equal to a specified home-count or home-frequency threshold; and
determining that: a total number or frequency of the second-type data points in the dataset is at least equal to a specified visitor-count or visitor-frequency threshold.
19. The method of claim 18 , further comprising:
selecting from the dataset a third-type data point comprising a second user identifier and
the second network; selecting from the dataset a fourth-type data point comprising the second user identifier and the first network address; and
increasing the baseline confidence level.
20. The method of claim 18 , further comprising:
increasing the baseline confidence level.
21. The method of claim 18 , further comprising:
receiving an enrichment attribute; testing whether the enrichment attribute is associated with the first entity or the second entity; and
increasing the baseline confidence level, if the enrichment attribute is determined to be associated with the first entity or the second entity; and
otherwise, decreasing the baseline confidence level or invalidating the inferred relationship.
22. A system for inferring a relationship between two entities, comprising:
a processor;
a network port in communication with the processor and adapted to receive from a data source a network observation dataset, each data point in the dataset comprising a user identifier and a network address; and
a memory coupled to the processor and comprising instructions, which when executed by the processor, program the processor to:
select from the dataset a first-type data point comprising a first user identifier and a first network address;
identify a first entity associated with the first network address;
select from the dataset a second-type data point comprising the first user identifier and a second network address different from the first network address;
identify a second entity that is different from the first entity and that is associated with the second network address;
infer at a baseline confidence level a relationship between the first entity and the second entity;
determine that: a total number or frequency of the first-type data points in the dataset is at least equal to a specified home-count or home-frequency threshold; and
determine that: a total number or frequency of the second-type data points in the dataset is at least equal to a specified visitor-count or visitor-frequency threshold.
23. The system of claim 22 , wherein the instructions further program the processor to:
select from the dataset a third-type data point comprising a second user identifier and the second network; select from the dataset a fourth-type data point comprising the second user identifier and the first network address; and increase the baseline confidence level.
24. The system of claim 22 , wherein the instructions further program the processor to increase the baseline confidence level.
25. The system of claim 22 , wherein the instructions further program the processor to:
receive an enrichment attribute;
test whether the enrichment attribute is associated with the first entity or the second entity; and
increase the baseline confidence level, if the enrichment attribute is determined to be associated with the first entity or the second entity; and
otherwise, decrease the baseline confidence level or invalidating the inferred relationship.