GENERATING HARD INSTANCES OF CAPTCHAS

• US 20100077209A1
• Filed: 09/24/2008
• Published: 03/25/2010
• Est. Priority Date: 09/24/2008
• Status: Abandoned Application

• Current Assignee: Oath Inc. (Verizon Communications Inc.)
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: RAVIKUMAR, Shanmugasundaram, BRODER, Andrei
• Application Number:
  US12/236,869
• Publication Number:
  US 20100077209A1
• Time in Patent Office: Days
• US Class Current: 713/168
• CPC Class Codes: G06F 21/46 by designing passwords or checking the strength of passwords

Abstract

Methods and systems are described for enhancing the difficulty of captchas and enlarging a core of available captchas that are hard for an automated or robotic user to crack.

Claims

1. A computer-implemented method for modifying a set of captchas based on responses to the captchas from one or more client computers, comprising:

classifying first ones of the responses as coming from an automated process and second ones of the responses as coming from a human;

modifying a first one of the captchas for which the first responses represent a corresponding success rate higher than a first threshold; and

eliminating a second one of the captchas from the set of captchas for which the second responses represent a corresponding failure rate above a second threshold.

2. The method of claim 1, further comprising adding new captchas determined to be difficult for an automated process but not for humans.

3. The method of claim 1, further comprising deriving the set of captchas from a larger group of captchas.

4. The method of claim 1, further comprising:

monitoring the use of a service by a user and determining if usage characteristics of the user are correlated with usage characteristics of an automated robotic user.

5. The method of claim 4, wherein usage characteristics comprise registration attributes, and wherein monitoring the use comprises monitoring registration attributes.

6. The method of claim 4 wherein usage characteristics comprise post registration usage of the service, and wherein monitoring the use comprises monitoring the post registration usage of the service.

7. A computer system for selectively accepting access requests to a service, the computer system configured to:

determine a hard set of captchas from a plurality of possible captchas;

render some or all of the hard set of captchas on a computing device;

receive responses to the rendered hard set of captchas;

track the received responses to the rendered hard set of captchas;

distinguish between responses believed to be entered by a human and responses believed to be entered by an automated client; and

eliminate a group of the hard set of captchas, the eliminated group having a failure rate of response above an acceptable threshold for those responses believed to be entered by a human.

8. The computer system of claim 7 wherein in order to distinguish between responses believed to be entered by a human and responses believed to be entered by an automated client the computer system is configured to determine if usage characteristics of the user are correlated with usage characteristics of an automated robotic user.

9. The computer system of claim 8, wherein usage characteristics comprise registration attributes, and wherein the computer system is configured to monitor registration attributes.

10. The computer system of claim 8, wherein usage characteristics comprise post registration usage of the service, and wherein the computer system is configured to monitor the post registration usage of the service.

11. A computer-implemented method for selectively accepting access requests from a client computer connected to a server computer, comprising:

presenting a plurality of captchas to a plurality of users wishing to access a service;

receiving answers to the captchas;

monitoring registration for the service by a user and determining if registration characteristics of the user are correlated with characteristics of a robotic user;

monitoring the post registration use of the service by a user and determining if post registration usage characteristics of the user are correlated with usage characteristics of a robotic user;

assessing the answers to the captchas and tracking correct and incorrect of the answers; and

classifying the captchas that receive incorrect answers from a suspected robotic user for inclusion in a hard set.

12. A computer-implemented method, comprising:

causing an original set of captchas to be rendered on a first plurality of client computers; and

causing a modified set of captchas to be rendered on a second plurality of client computers, the modified set of captchas including a modified captcha corresponding to a first captcha from the original set of captchas, the modified captcha having been modified as a result of responses to the first captcha by automated processes, the modified captcha being more difficult for the automated processes to successfully process.