Detecting click spam

• US 8,694,374 B1
• Filed: 03/14/2007
• Issued: 04/08/2014
• Est. Priority Date: 03/14/2007
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Diligenti, Michelangelo, Federici, Diego
• Primary Examiner(s): Weiss, John G
• Assistant Examiner(s): Sigmond, Bennett
• Application Number:
  US11/686,288
• Time in Patent Office: 2,582 Days
• Field of Search: 705/14.47
• US Class Current: 705/14.47
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

A computer-implemented method for processing network activities is described. The method includes identifying a model that specifies attributes for network objects, identifying a network object having one or more attributes that deviate from the model, and providing as an input to a ranking algorithm a value associated with the deviance of the one or more attributes of the identified network object.

Claims

1. A computer-implemented method, the method comprising:

obtaining a model of typical behavior for each of a plurality of different statistics derived from selection data associated with each of a plurality of user identifiers, the selection data representing prior selections of search results by users associated with respective user identifiers, wherein the search results are responsive to respective search queries;

determining a distribution of counts of deviations to counts of the user identifiers having those deviations, wherein a particular deviation is a calculated statistic for a user identifier that deviates from typical behavior for the statistic according to the model;

determining a plurality of zones within the distribution of counts of deviations to counts of the user identifiers having those deviations, wherein each zone defines a respective range of deviations in the distribution;

determining that a first user identifier of the plurality of user identifiers has a count of deviations that is within a particular zone of the plurality of zones within the distribution; and

modifying, by one or more computers, selection data associated with the first user identifier based on a weight associated with the particular zone within the distribution in response to determining that the first user identifier has a count of deviations that is within the particular zone within the distribution.

2. The method of claim 1 wherein a user identifier is an Internet Protocol address, a cookie, a media access control address, or a hardware configuration for a client device.

3. The method of claim 1 wherein modifying selection data associated with the first user identifier based on a weight associated with the particular zone within the distribution comprises reducing an influence of one or more attributes of selection data associated with the first user according to the weight when computing statistics between search queries and search results.

4. The method of claim 1, wherein a calculated statistic in the plurality of different statistics is a click to query ratio, a count of click types for different search results, a count of clicked search results per total number of clicks, a count of queries per number of distinct queries, a median click length, a maximum number of clicks per time period, a distribution of the click lengths, a distribution of click positions, a fraction of clicks outside a first page of search results, or a fraction of clicks on advertisements.

5. The method of claim 4 wherein a click type in the click types is a duration of the click.

6. The method of claim 1 wherein the typical behavior for the statistic according to the model is based at least partly on a prototypical distribution for the statistic.

7. The method of claim 6 wherein the prototypical distribution for the statistic is an average of values of the statistic for the plurality of user identifiers.

8. The method of claim 1, wherein modifying selection data associated with the first user identifier based on a weight associated with the particular zone within the distribution comprises:

providing the weight to a ranking process for use in discounting the selection data associated with the first user identifier.

9. The method of claim 1, wherein modifying selection data associated with the first user identifier based on a weight associated with the particular zone within the distribution comprises discounting one or more selection statistics associated with the first user identifier according to the weight.

10. The method of claim 1, further comprising:

calculating the plurality of different statistics; and

building the model of typical behavior for each statistic for the plurality of user identifiers.

11. The method of claim 1, wherein modifying selection data associated with the first user identifier comprises limiting a contribution of selection data associated with the first user identifier.

12. The method of claim 1, further comprising:

providing the modified selection data to a ranking process for ranking search results responsive to search queries.

13. A system comprising:

one or more data processing apparatus; and

a computer-readable storage device having instructions stored thereon that, when executed by the one or more data processing apparatus, cause the one or more data processing apparatus to perform operations comprising;

obtaining a model of typical behavior for each of a plurality of different statistics derived from selection data associated with each of a plurality of user identifiers, the selection data representing prior selections of search results by users associated with respective user identifiers, wherein the search results are responsive to respective search queries;

determining a distribution of counts of deviations to counts of the user identifiers having those deviations, wherein a particular deviation is a calculated statistic for a user identifier that deviates from typical behavior for the statistic according to the model;

determining a plurality of zones within the distribution of counts of deviations to counts of the user identifiers having those deviations, wherein each zone defines a respective range of deviations in the distribution;

determining that a first user identifier of the plurality of user identifiers has a count of deviations that is within a particular zone of the plurality of zones within the distribution; and

modifying selection data associated with the first user identifier based on a weight associated with the particular zone within the distribution in response to determining that the first user identifier has a count of deviations that is within the particular zone within the distribution.

14. The system of claim 13, wherein a user identifier is an Internet Protocol address, a cookie, a media access control address, or a hardware configuration for a client device.

15. The system of claim 13, wherein modifying selection data associated with the first user identifier based on a weight associated with the particular zone within the distribution comprises reducing an influence of one or more attributes of selection data associated with the first user according to the weight when computing statistics between search queries and search results.

16. The system of claim 13, wherein a calculated statistic in the plurality of different statistics is a click to query ratio, a count of click types for different search results, a count of clicked search results per total number of clicks, a count of queries per number of distinct queries, a median click length, a maximum number of clicks per time period, a distribution of the click lengths, a distribution of click positions, a fraction of clicks outside a first page of search results, or a fraction of clicks on advertisements.

17. The system of claim 16, wherein a click type in the click types is a duration of the click.

18. The system of claim 13, wherein the typical behavior for the statistic according to the model is based at least partly on a prototypical distribution for the statistic.

19. The system of claim 18, wherein the prototypical distribution for the statistic is an average of values of the statistic for the plurality of user identifiers.

20. The system of claim 13, wherein modifying selection data associated with the first user identifier based on a weight associated with the particular zone within the distribution comprises:

providing the weight to a ranking process for use in discounting the selection data associated with the first user identifier.

21. The system of claim 13, wherein modifying selection data associated with the first user identifier based on a weight associated with the particular zone within the distribution comprises discounting one or more selection statistics associated with the first user identifier according to the weight.

22. The system of claim 13, wherein the operations further comprise:

calculating the plurality of different statistics; and

building the model of typical behavior for each statistic for the plurality of user identifiers.

23. The system of claim 13, wherein modifying selection data associated with the first user identifier based on a weight associated with the particular zone within the distribution comprises limiting a contribution of selection data associated with the first user identifier.

24. The system of claim 13, wherein the operations further comprise:

providing the modified selection data to a ranking process for ranking search results responsive to search queries.