User identification across social media

Patent metadata

Field	Value
Publication number	US-9544381-B2
Application number	US-201414210372-A
Country	US
Kind code	B2
Filing date	Mar 13, 2014
Priority date	Mar 13, 2013
Publication date	Jan 10, 2017
Grant date	Jan 10, 2017

Abstract

Users may be identified across websites, such as social media websites. Prior user information data and candidate user information data may be received. An algorithm may identify a first plurality of behavioral patterns in the prior user information data and a second plurality of behavioral patterns in the candidate user information datum. The algorithm may determine whether the candidate user information datum and the prior user information data correspond to the same user based, at least in part, on the first and second pluralities of behavioral patterns.

First claim

What is claimed is:

1. A method, comprising:
receiving prior user information data comprising a plurality of usernames corresponding to a known user;
receiving a candidate user information datum comprising a username corresponding to an unknown user;
identifying a first behavioral pattern in the prior user information data based on content in the plurality of usernames, wherein the first behavioral pattern identifies the known user based on the content in the plurality of usernames, and a second behavioral pattern based on a content in the username in the candidate user information datum, wherein the second behavioral pattern is based on the content in the username and is indicative of the unknown user,
wherein the step of identifying a first behavioral pattern in the prior user information data based on content in the plurality of usernames comprises the step of determining a distribution of dynamic time warping (DTW) distances for the plurality of usernames, and wherein the step of determining whether the candidate user information datum and the prior user information data correspond to the same user comprises determining whether a distribution of DTW distances indicates that the candidate user information datum belongs to the known user; and
determining, based at least in part on the first and second behavioral patterns, whether the candidate user information datum and the prior user information data correspond to a user such that the known user and the unknown user are the same user.

2. The method of claim 1 , wherein the first behavioral pattern is one of a first plurality of behavioral patterns, and wherein the second behavioral pattern is one of a second plurality of behavioral patterns.

3. The method of claim 2 , further comprising:
determining redundant information data based on the first plurality of behavioral patterns;
constructing a plurality of data features, wherein each data feature of the plurality of data features corresponds to a redundant information datum of the redundant information data; and
generating a prediction model based on the constructed plurality of data features, wherein determining whether the candidate user information datum and the prior user information data correspond to the same user is further based on the prediction model.

4. The method of claim 2 , further comprising computing statistical properties of the prior user information data and the candidate user information datum based on the first and second pluralities of behavioral patterns, wherein determining whether the candidate user information datum and the prior user information data correspond to the same user is further based on the statistical properties.

5. The method of claim 2 , wherein the prior user information data comprises user information data on a plurality of websites, and the first plurality of behavioral patterns are identified based on the user information data on the plurality of websites.

6. The method of claim 2 , wherein the first and second pluralities of behavioral patterns comprise patterns based on at least one of: human limitation, exogenous factors, and endogenous factors.

7. A computer program product, comprising:
a non-transitory computer readable medium comprising code for performing the steps of:
receiving prior user information data comprising a plurality of usernames corresponding to a known user;
receiving a candidate user information datum comprising a username corresponding to an unknown user;
identifying a first behavioral pattern in the prior user information data based on content in the plurality of usernames, wherein the first behavioral pattern identifies the known user based on the content in the plurality of usernames, and a second behavioral pattern based on a content in the username in the candidate user information datum, wherein the second behavioral pattern is based on the content in the username and is indicative of the unknown user,
wherein the step of identifying a first behavioral pattern in the prior user information data based on content in the plurality of usernames comprises the step of determining a distribution of dynamic time warping (DTW) distances for the plurality of usernames, and wherein the step of determining whether the candidate user information datum and the prior user information data correspond to the same user comprises determining whether a distribution of DTW distances indicates that the candidate user information datum belongs to the known user; and
determining, based at least in part on the first and second behavioral patterns, whether the candidate user information datum and the prior user information data correspond to a user such that the known user and the unknown user are the same user.

8. The computer program product of claim 7 , wherein the first behavioral pattern is one of a first plurality of behavioral patterns, and wherein the second behavioral pattern is one of a second plurality of behavioral patterns.

9. The computer program product of claim 8 , wherein the medium further comprises code for performing the steps of:
determining redundant information data based on the first plurality of behavioral patterns;
constructing a plurality of data features, wherein each data feature of the plurality of data features corresponds to a redundant information datum of the redundant information data; and
generating a prediction model based on the constructed plurality of data features, wherein determining whether the candidate user information datum and the prior user information data correspond to the same user is further based on the prediction model.

10. The computer program product of claim 8 , wherein the medium further comprises code for performing the step of computing statistical properties of the prior user information data and the candidate user information datum based on the first and second pluralities of behavioral patterns, wherein determining whether the candidate user information datum and the prior user information data correspond to the same user is further based on the statistical properties.

11. The computer program product of claim 8 , wherein the prior user information data comprises user information data on a plurality of websites, and the first plurality of behavioral patterns are identified based on the user information data on the plurality of websites.

12. The computer program product of claim 8 , wherein the first and second pluralities of behavioral patterns comprise patterns based on at least one of: human limitation, exogenous factors, and endogenous factors.

13. An apparatus, comprising:
a memory; and
a processor coupled to the memory, wherein the processor is configured to execute the steps of:
receiving prior user information data comprising a plurality of usernames corresponding to a known user;
receiving a candidate user information datum comprising a username corresponding to an unknown user;
identifying a first behavioral pattern in the prior user information data based on content in the plurality of usernames, wherein the first behavioral pattern identifies the known user based on the content in the plurality of usernames, and a second behavioral pattern based on a content in the username in the candidate user information datum, wherein the second behavioral pattern is based on the content in the username and is indicative of the unknown user,
wherein the step of identifying a first behavioral pattern in the prior user information data based on content in the plurality of usernames comprises the step of determining a distribution of dynamic time warping (DTW) distances for the plurality of usernames, and wherein the step of determinin

Assignees

• Zafarani Reza
• Liu Huan
• Univ Arizona State

Inventors

• Zafarani Reza
• Liu Huan

Classifications

• H04L67/22 Primary

  Electricity · mapped topic

• H04L67/535 Primary

  Tracking the activity of the user (network monitoring arrangements H04L43/00; recording of computer activity G06F11/34) · CPC title