LOCATION ESTIMATION OF SOCIAL NETWORK USERS

• US 20130325977A1
• Filed: 08/24/2012
• Published: 12/05/2013
• Est. Priority Date: 06/04/2012
• Status: Active Grant

• Current Assignee: Snap, Inc.
• Original Assignee: International Business Machines Corporation
• Inventors: DREWS, Clemens, Mahmud, Jalal U., Nichols, Jeffrey W.
• Granted Patent: US 9,002,960 B2
• Time in Patent Office: Days
• US Class Current: 709/206
• CPC Class Codes: G06F 16/00 Information retrieval; Database structures therefor; File system structures therefor
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 50/01 Social networking
• CPC Class Codes: H04L 51/52 for supporting social networking services

Abstract

Various embodiments relate to estimating the location of social network users. In one embodiment, a plurality of social media messages generated by a given user is received. A plurality of location features is extracted from the social media messages. Each of the location features is processed with at least one classifier from an ensemble of classifiers. A location classification is generated by each of the classifiers for each of the social media messages. Each classification comprises a location and a weight associated with that location. Once of the locations is selected from the location classifications as the location of the given user based on a combination of the weights of the location classifications.

Claims

1. A method comprising:

receiving a plurality of social media messages generated by a given user;

extracting a plurality of location features from the social media messages;

processing each of the location features with at least one classifier from an ensemble of classifiers;

generating, by each of the classifiers, a location classification for each of the social media messages, each location classification comprising a location and a weight associated with that location; and

selecting one of the locations from the location classifications as the location of the given user based on a combination of the weights of the location classifications.

2. The method of claim 1,

wherein the ensemble of classifiers is a hierarchical ensemble of classifiers, and

each classifier in a given level of the hierarchical ensemble of classifiers identifies a location at a lower granularity than each classifier in a higher level of the hierarchical ensemble of classifiers.

3. The method of claim 2, wherein the generating comprises:

identifying, by each classifier at a first level of the hierarchical ensemble of classifiers, a time-zone location for the given user; and

identifying, by each classifier at a second level of the hierarchical ensemble of classifiers that is below the first level, a city location for the given user based partially on the time-zone location identified in the first level.

4. The method of claim 2, wherein the generating comprises:

identifying, by each classifier at a first level of the hierarchical ensemble of classifiers, a state location for the given user; and

identifying, by each classifier at a second level of the hierarchical ensemble of classifiers that is below the first level, a city location for the given user based partially on the state location identified in the first level.

5. The method of claim 2, wherein the generating comprises:

identifying, by each classifier at a first level of the hierarchical ensemble of classifiers, a geographical region location associated with the given user; and

identifying, by each classifier at a second level of the hierarchical ensemble of classifiers that is below the first level, a city location for the given user based partially on the geographical region location identified in the first level.

6. The method of claim 1, wherein the processing comprises processing subsets of the location features with at least one statistical classifier, the subsets of location features comprising terms that are local to a given city, metadata tags, and place names.

7. The method of claim 1, wherein the processing comprises processing a subset of the location features with at least one heuristic classifier, the subset of location features comprising at least one of a frequency of city names and a frequency of state names in the social media messages.

8. The method of claim 1, wherein the processing comprises processing a subset of the location features with at least one heuristic classifier, the subset of location features comprising a frequency of visited locations that are identified from the social media messages based on uniform resource locators within the social media messages.

9. The method of claim 1, wherein the processing comprises processing a subset of the location features with at least one behavior-based classifier, the subset of location features comprising a frequency of social media messages generated by the given user for a plurality of time durations.

10. The method of claim 1, wherein the processing comprises:

determining, by a binary classifier associated with one of the classifiers in the ensemble of classifiers, if a location associated with a given user is predictable by the one classifier; and

preventing the one classifier from generating the location classification if the binary classifier determines that the location is not predictable by the one classifier.

11. The method of claim 1, wherein the processing comprises:

processing at least one subset of the location features with at least one statistical classifier;

processing at least one subset of the location features with at least one heuristic classifier; and

processing at least one subset of the location features with at least one behavior-based classifier.

12. The method of claim 11, wherein the processing further comprises determining, by a binary classifier associated with one of the classifiers in the ensemble of classifiers, if a location associated with a given user is predictable by the one classifier.