Machine learning and validation of account names, addresses, and/or identifiers

• US 9,928,465 B2
• Filed: 05/20/2014
• Issued: 03/27/2018
• Est. Priority Date: 05/20/2014
• Status: Active Grant

• Current Assignee: Yahoo Assets LLC
• Original Assignee: Oath Inc. (Verizon Communications Inc.)
• Inventors: McQueen, Donald J., Maxwell, Lachlan A.
• Primary Examiner(s): Brown, Sheree
• Application Number:
  US14/282,097
• Publication Number:
  US 20150339583A1
• Time in Patent Office: 1,407 Days
• Field of Search: 706 12
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: H04L 51/212 using filtering or selective blocking
• CPC Class Codes: H04L 63/0236 Filtering by address, protocol, port number or service, e.g. IP-address or URL
• CPC Class Codes: H04L 63/0245 Filtering by information in the payload
• CPC Class Codes: H04L 63/126 the source of the received data
• CPC Class Codes: H04L 63/1425 Traffic logging, e.g. anomaly detection
• CPC Class Codes: H04L 63/1466 Active attacks involving interception, injection, modification, spoofing of data unit addresses, e.g. hijacking, packet injection or TCP sequence number attacks

Abstract

Systems and methods are disclosed for determining if an account identifier is computer-generated. One method includes receiving the account identifier, dividing the account identifier into a plurality of fragments, and determining one or more features of at least one of the fragments. The method further includes determining the commonness of at least one of the fragments, and determining if the account identifier is computer-generated based on the features of at least one of the fragments, and the commonness of at least one of the fragments.

Claims

1. A computer-implemented method for determining if an account identifier is computer-generated, comprising:

receiving the account identifier;

dividing the account identifier into a plurality of fragments;

calculating a first feature of the plurality of fragments of the account identifier by calculating fragment frequencies of each of the plurality of fragments relative to a plurality of fragments associated with a plurality of account identifiers;

calculating a second feature of the account identifier by classifying alphanumeric character types of the account identifier;

calculating a third feature of the plurality of fragments of the account identifier by hashing the fragment frequencies to a commonness of each of the plurality of fragments of the account identifier; and

generating a plurality of data pairs, each data pair comprising a fragment of the plurality of fragments and an associated feature value, the associated feature value comprising the first feature, the second feature, or the third feature;

providing each of the plurality of data pairs to a probabilistic classifier model to determine if the account identifier is computer-generated.

2. The method of claim 1, wherein, in response to determining that the at least one of the fragments of the account identifier cannot be found in the plurality of account identifiers, increasing a probability that the account identifier is computer-generated.

3. The method of claim 1, wherein the probabilistic classifier model is trained with a training set of a plurality of account identifiers that are known to be or known to not be computer-generated, and wherein the training set allows the probabilistic classifier model to independently weigh the first feature, second feature, and third feature in order to more accurately determine if the account identifier is computer-generated.

4. The method of claim 1, further comprising:

calculating one or more features of the account identifier by counting characters by character type.

5. The method of claim 1, wherein at least one of the plurality of fragments is truncated to contain only consonants.

6. The method of claim 1, wherein each character of at least one of the plurality of fragments is hashed according to character type.

7. The method of claim 6, wherein each character type is selected from a group including consonant, vowel, number, and punctuation mark.

8. A system for determining if an account identifier is computer-generated, the system including:

a data storage device storing instructions determining if an account identifier is computer-generated; and

a processor configured to execute the instructions to perform a method including;

receiving the account identifier;

dividing the account identifier into a plurality of fragments;

calculating a first feature of the plurality of fragments of the account identifier by calculating fragment frequencies of each of the plurality of fragments relative to a plurality of fragments associated with a plurality of account identifiers;

calculating a second feature of the account identifier by classifying alphanumeric character types of the account identifier;

calculating a third feature of the plurality of fragments of the account identifier by hashing the fragment frequencies to a commonness of each of the plurality of fragments of the account identifier; and

generating a plurality of data pairs, each data pair comprising a fragment of the plurality of fragments and an associated feature value, the associated feature value comprising the first feature, the second feature, or the third feature;

providing each of the plurality of data pairs to a probabilistic classifier model to determine if the account identifier is computer-generated.

9. The system of claim 8, wherein, in response to determining that the at least one of the fragments of the account identifier cannot be found in the plurality of account identifiers, increasing a probability that the account identifier is computer-generated.

10. The system of claim 8, wherein the probabilistic classifier model is trained with a training set of a plurality of account identifiers that are known to be or known to not be computer-generated, and wherein the training set allows the probabilistic classifier model to independently weigh the first feature, second feature, and third feature in order to more accurately determine if the account identifier is computer-generated.

11. The system of claim 8, wherein the processor is further configured for:

calculating one or more features of the account identifier by counting characters by character type.

12. The system of claim 8, wherein at least one of the plurality of fragments is truncated to contain only consonants.

13. The system of claim 8, wherein each character of at least one of the plurality of fragments is hashed according to character type.

14. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform a method for determining whether an account identifier is computer-generated, the method including:

receiving the account identifier;

dividing the account identifier into a plurality of fragments;

calculating a first feature of the plurality of fragments of the account identifier by calculating fragment frequencies of each of the plurality of fragments relative to a plurality of fragments associated with a plurality of account identifiers;

calculating a second feature of the account identifier by classifying alphanumeric character types of the account identifier;

calculating a third feature of the plurality of fragments of the account identifier by hashing the fragment frequencies to a commonness of each of the plurality of fragments of the account identifier; and

generating a plurality of data pairs, each data pair comprising a fragment of the plurality of fragments and an associated feature value, the associated feature value comprising the first feature, the second feature, or the third feature;

providing each of the plurality of data pairs to a probabilistic classifier model to determine if the account identifier is computer-generated.

15. The non-transitory computer-readable medium of claim 14, wherein, in response to determining that the at least one of the fragments of the account identifier cannot be found in the plurality of account identifiers, increasing a probability that the account identifier is computer-generated.

16. The non-transitory computer-readable medium of claim 14, wherein the probabilistic classifier model is trained with a training set of a plurality of account identifiers that are known to be or known to not be computer-generated, and wherein the training set allows the probabilistic classifier model to independently weigh the first feature, second feature, and third feature in order to more accurately determine if the account identifier is computer-generated.

17. The non-transitory computer-readable medium of claim 14, wherein each character of at least one of the plurality of fragments is hashed according to character type.