System and method for identifying contacts of a target user in a social network

• US 9,646,245 B2
• Filed: 10/29/2013
• Issued: 05/09/2017
• Est. Priority Date: 10/29/2012
• Status: Active Grant

• Current Assignee: Cognyte Technologies Israel Ltd. (Cognyte Software Ltd.)
• Original Assignee: Verint Systems Limited (Verint Systems Incorporated)
• Inventors: Puzis, Rami, Stern, Roni, Rokach, Lior, Elovici, Yuval, Beja, Tal, Felner, Ariel, Bnaya, Zahy, Samama, Liron
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): Zidanic, Michael
• Application Number:
  US14/065,505
• Publication Number:
  US 20140229406A1
• Time in Patent Office: 1,288 Days
• Field of Search: None
• CPC Class Codes: G06F 16/2379 Updates performed during online database operations; commit processing
• CPC Class Codes: G06F 16/24575 using context
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/9536 Search customisation based on social or collaborative filtering
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/01 Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
• CPC Class Codes: G06Q 50/01 Social networking
• CPC Class Codes: H04L 67/306 User profiles

Abstract

When using Web intelligence (“Webint”) to collect information regarding a target social network user, one of the most valuable pieces of information is the target user'"'"'s List-Of-Friends (LOF). In some cases, however, the LOF of the target user is not accessible in his profile. Herein are described methods and systems for identifying the LOF of a target user. An analysis system crawls the profiles of social network users, other than the target user, and reconstructs the LOF of the target user from the crawled profiles.

Claims

1. A method, comprising:

receiving a request to identify a group of users of one or more social networks, who are defined as contacts of a target user of the social networks, wherein the target user'"'"'s contacts are hidden or otherwise inaccessible on the target user'"'"'s one or more social network profiles as a result of a privacy policy set by the target user;

responsively to the request, initially receiving a subset of the contacts of the target user and iteratively expanding the subset by analyzing only profiles of potential contacts that are likely to be contacts of the target user, wherein the analyzing comprises;

selecting a potential contact of the target user from among currently-known profiles based on a likelihood that the potential contact is a contact of the target user, the likelihood determined by evaluating a plurality of heuristics,

identifying that the potential contact is connected to the target user,

retrieving the contacts of the potential contact, and

adding the profiles of the contacts of the potential contact to the currently-known profiles.

2. The method according to claim 1, and comprising iteratively selecting the potential contact, identifying whether the potential contact is connected to the target user and updating the currently-known profiles, until meeting a termination condition that is based on the one or more social network'"'"'s response to the analysis.

3. The method according to claim 1, wherein the plurality of heuristics comprises a similarity in attribute values between the profile of the target user and the potential contact.

4. The method according to claim 1, wherein the plurality of heuristics comprises the number of other contacts of the target user that a potential contact is connected to.

5. The method according to claim 1, wherein the evaluating a plurality of heuristics comprises defining first and second heuristics, training a classifier to optimally combine the first and second heuristics using training data, and determining the likelihood based on the optimally-combined heuristics.

6. The method according to claim 5, wherein training the classifier comprises testing a default heuristic over a given number of initial potential contacts of the target user to produce training results, and combining the first and second heuristics based on the training results.

7. The method according to claim 5, wherein training the classifier comprises combining the first and second heuristics based on one or more previous searches for groups of users of one or more social networks, who are defined as contacts of other target users.

8. The method according to claim 1, wherein the plurality of heuristics comprises a likelihood of the potential contact to be connected to a contact of the target user, wherein the contact of the target user is connected to other contacts of the target user.

9. An apparatus, comprising:

an interface, which is configured to communicate with one or more social networks; and

a processor, which is configured to;

receive a request to identify a group of users of the social networks who are defined as contacts of a target user of the social networks, wherein the target user'"'"'s contacts are hidden or otherwise inaccessible on the target'"'"'s user'"'"'s one or more social network profiles as a result of a privacy policy set by the target user,

receive a subset of the contacts of the target user

iteratively expand the subset by analyzing only profiles of potential contacts that are likely to be contacts of the target user, wherein to analyze the profiles, the processor is configured to;

select a potential contact of the target user from among currently-known profiles based on a likelihood that the potential contact is a contact of the target user, the likelihood determined by evaluating a plurality of heuristics,

identify that the potential contact is connected to the target user, and

retrieve the contacts of the potential contact and add the profiles of the contacts of the potential contact to the currently-known profiles.

10. The apparatus according to claim 9, wherein the processor is configured to iteratively select the potential contact, identify whether the potential contact is connected to the target user and update the currently-known profiles, until meeting a termination condition that is based on the one or more social network'"'"'s response to the analysis.

11. A method for reconstructing a list of friends for a target user of a social network, the method comprising:

receiving a set of initial leads for the target user, the profile of the target user having a hidden list of friends;

obtaining a list of friends for each initial lead;

adding the profiles of the friends of each initial lead to a set of potential leads;

calculating the likelihood that each of the potential leads is a friend of the target user;

prioritizing the potential leads based on the likelihood;

selecting a potential lead having the highest priority;

analyzing the selected potential lead'"'"'s list of friends;

determining, based on the analysis, that the selected potential lead is a friend of target user or not a friend of the target user, wherein for a selected potential lead that is a friend of the target user;

moving the selected potential lead from the set of potential leads to the target user'"'"'s list of friends,

adding all friends of the selected potential lead to the set of potential leads, and

updating a currently-known graph (CKG) describing the relationships of currently-known profiles in the social network;

repeating the calculating, prioritizing, selecting, analyzing, and determining for a fixed number of iterations; and

presenting the target user'"'"'s reconstructed list of friends, wherein the reconstructed list of friends is at least a portion of the target user'"'"'s hidden list of friends.