Automatic prediction of aspects of human task performance

• US 9,026,483 B1
• Filed: 11/19/2007
• Issued: 05/05/2015
• Est. Priority Date: 11/19/2007
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Cohen, Peter D.
• Primary Examiner(s): Hill, Stanley K
• Application Number:
  US11/942,411
• Time in Patent Office: 2,724 Days
• Field of Search: None
• US Class Current: 706/46
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06Q 10/06375 Prediction of business process outcome or impact based on a proposed change

Abstract

Techniques are described for automatically facilitating the performance of tasks based on automatically generating predictions about future task performance, and then using the generated task performance prediction information in various ways. For example, as tasks are performed via a task exchange server system, various information about the tasks and their performance may be tracked, and the tracked information may then be analyzed to identify relationships between various of the information related to the tasks and to corresponding information about the performance of the tasks. Based on the identified relationship information, the task exchange server may automatically generate predictions regarding how performance of additional tasks will occur in the future, such as by determining task performance information that corresponds to information available about the additional tasks. The automatically generated task performance prediction information may then be used to facilitate performance of tasks in various manners.

Claims

1. A computer-implemented method for a task exchange server to automatically facilitate performance of tasks by human task performer users based on automated predictions of task performance parameters related to the performance of the tasks, the method comprising:

automatically facilitating performance of a plurality of tasks of multiple types that are submitted to the task exchange server by multiple task requesters, the plurality of tasks being performed by a mutable collection of multiple human task performer users of the task exchange server who are unaffiliated with the task requesters, the automatic facilitating of the performance of each task being performed by one or more configured computing systems of the task exchange server and including obtaining results of the performance of the task by one or more human task performer users who select to perform the task and providing the obtained performance results to the task requester that submitted the task;

tracking information about the performance of the plurality of tasks, the tracked information including information about attributes of the plurality of tasks that affect their performance based at least in part by being discernible by task performer users who select whether to perform the tasks, the tracked information further including information about values for performance parameters that each measure one or more aspects of the performance of the tasks;

after the performance of the plurality of tasks, automatically analyzing the tracked information in order to determine correlations for each of one or more identified attributes to an identified value for at least one performance parameter, such that the performance of a task having an identified attribute is likely to result in the identified at least one performance parameter value that is correlated to that identified attribute, the automatic analyzing being performed by one or more configured computing systems;

receiving information from a first task requester indicating one or more new tasks that have not yet been performed and that are distinct from the plurality of tasks, the one or more new tasks having indicated attributes that include at least one of the identified attributes;

in response to the received information, automatically predicting likely values for one or more performance parameters that will occur during a possible future performance of the indicated tasks, the automatic predicting being performed by one or more configured computing systems and being based at least in part on the at least one attributes of the indicated new tasks and on the determined correlations, the likely performance parameter values for the possible future performance being predicted without an identification of which human task performer users will select to participate in the possible future performance;

providing the predicted likely performance parameter values to the first task requester; and

after the providing of the predicted likely performance parameter values and before performance of the indicated new tasks, receiving from the first task requester one or more submitted tasks to be performed by task performer users, the one or more submitted tasks being an altered version of the indicated new tasks that have one or more altered attributes differing from the indicated attributes, the altered attributes affecting the values that will likely occur for the one or more performance parameters during future performance of the one or more submitted tasks, to enable the first task requester to have enhanced control over performance of tasks.

2. The method of claim 1 wherein the one or more altered attributes include at least one of a specification by the first task requester of when the one or more submitted tasks are to be performed, a specification by the first task requester of a manner in which performance of the one or more submitted tasks is to occur, and a specification by the first task requester of a price to be paid to one or more of the human task performer users to perform the one or more submitted tasks, wherein the one or more performance parameters measure at least one of an amount of time until the one or more submitted tasks will be performed and a quality of results of performance of the one or more submitted tasks, and wherein the method further comprises facilitating performance of the one or more submitted tasks by the one or more human task performer users and providing the results of the performance to the first task requester, the performance resulting in actual values for the one or more performance parameters that differ from the predicted likely performance parameter values.

3. A computer-implemented method for automatically facilitating performance of tasks by human task performer users based on predicted task performance parameters, the method comprising:

tracking information about performance of a plurality of tasks by multiple human task performer users via a task exchange server, the plurality of tasks having been supplied to the task exchange server for performance by multiple task requesters unaffiliated with the multiple human task performer users;

automatically analyzing the tracked information in order to identify correlations between attributes of the plurality of tasks and performance parameters for the plurality of tasks, the attributes of the plurality of tasks reflecting information about the tasks that is available to human task performer users who select whether to perform the tasks, the performance parameters for the plurality of tasks each corresponding to one or more aspects of the performance of the plurality of tasks, the automatic analyzing being performed by one or more configured computing systems;

receiving an indication of one or more tasks that are not yet performed and that have one or more indicated attributes, the indicated tasks being distinct from the plurality of tasks;

after the automatic analyzing of the track information about the performance of the plurality of tasks, automatically predicting likely information about one or more performance parameters for a future performance of the indicated tasks, the automatic predicting being performed by one or more configured computing systems and being based at least in part on the one or more indicated attributes and on the identified correlations from the automatic analyzing; and

before performance of the indicated tasks has occurred, providing the predicted likely information about the one or more performance parameters for the indicated tasks, to enable the future performance of the indicated tasks to be altered based on modifications to at least one of the indicated attributes of the indicated tasks.

4. The method of claim 3 wherein the received indication of the one or more tasks is from one of the multiple task requesters, and wherein the providing of the predicted likely information includes providing the predicted likely information to the one task requester.

5. The method of claim 4 further comprising, after the providing of the predicted likely information to the one task requester, receiving from the one task requester one or more supplied tasks to be performed by one or more of the multiple human task performer users, the one or more supplied tasks being an altered version of the indicated tasks so as to have one or more modified attributes that differ from the indicated attributes, the modified attributes affecting performance parameter values that will likely occur during future performance of the one or more supplied tasks.

6. The method of claim 5 further comprising, under control of the one task requester:

receiving the provided predicted likely information about the one or more performance parameters;

determining that the predicted likely information for at least one of the one or more performance parameters is not preferred;

determining to modify one or more of the indicated attributes for the indicated tasks in such a manner as to alter likely information for the at least one performance parameter; and

providing the one or more supplied tasks for performance with the modified one or more attributes.

7. The method of claim 4 wherein the received indication of the one or more tasks from the one task requester is part of a request from the one task requester for the predicted likely information about the one or more performance parameters, the request being made by the one task requester prior to the one task requester submitting the one or more tasks to the task exchange server for performance, the request including an indication of the one or more performance parameters for which the likely information is to be predicted for the future performance.

8. The method of claim 3 wherein the received indication of the one or more tasks includes an indication of a group of multiple tasks of multiple types that are selected to represent tasks that may be received by the task exchange server from multiple task requesters during a future period of time, and wherein the group of multiple tasks is selected by the task exchange server to assess likely behavior of the task exchange server in facilitating performance of the multiple tasks of the group.

9. The method of claim 3 wherein the received indication of the one or more tasks with the one or more indicated attributes is part of receiving multiple groups of alternative attributes for the indicated tasks, wherein the automatic predicting of the likely information about the one or more performance parameters for a future performance of the indicated tasks is performed for each of the multiple groups of alternative attributes so as to generate differing predicted likely information for at least some of the multiple groups of alternative attributes, and wherein the providing of the predicted likely information is performed for each of the multiple groups of alternative attributes, so as to illustrate how the information about the one or more performance parameters is likely to change as attributes for the indicated tasks change.

10. The method of claim 3 wherein the providing of the predicted likely information includes providing the predicted likely information to the task exchange server, and wherein the method further comprises, after the providing, automatically modifying one or more of the indicated attributes for the indicated tasks in such a manner as to alter likely information for at least one of the one or more performance parameters, and facilitating performance of the one or more indicated tasks with the modified attributes by one or more of the multiple human task performer users.

11. The method of claim 3 wherein the one or more performance parameters each measure one or more aspects of performance of tasks, the one or more aspects including an amount of time until the one or more submitted tasks are performed.

12. The method of claim 3 wherein the one or more performance parameters each measure one or more aspects of performance of tasks, the one or more aspects including an amount of money to be paid for performance of a task so as to obtain performance within an indicated period of time and/or to obtain results of performance that satisfy an indicated measure of quality.

13. The method of claim 3 wherein the one or more performance parameters each measure one or more aspects of performance of tasks, the one or more aspects including a quality of results of performance of a task that are obtained.

14. The method of claim 3 wherein the predicted likely information about the one or more performance parameters include one or more predicted likely values for each of the one or more performance parameters.

15. The method of claim 3 wherein the one or more indicated attributes reflect at least one of a specification of a deadline for when the one or more tasks are to be performed, and a specification of a price to be paid to one or more of the human task performer users to perform the one or more tasks.

16. The method of claim 3 wherein the one or more indicated attributes reflect at least one of a category of the one or more tasks, a type of the one or more tasks, a type of information to be analyzed as part of performance of the one or more tasks, one or more qualifications needed by task performer users who perform the one or more tasks, a task requester who supplied the one or more tasks, one or more types of information supplied by the task requester to specify how the one or more tasks are to be performed, and an indicated minimum level of quality of results of the performance of the one or more tasks.

17. The method of claim 3 wherein the analyzing of the tracked information further identifies additional correlations between characteristics of human task performer users who were available to perform the plurality of tasks and the performance parameters for the plurality of tasks, and wherein the automatic predicting of the likely information about the one or more performance parameters for the future performance of the indicated tasks is further based at least in part on information about characteristics of human task performer users who are expected to be available for the future performance of the indicated tasks and on the identified additional correlations.

18. A non-transitory computer-readable medium whose contents configure a computing system to perform a method to automatically facilitate performance of tasks by task performer users based on predicted task performance parameters, the method comprising:

automatically analyzing information about prior performance of a plurality of tasks in order to identify relationships between attributes corresponding to the plurality of tasks and performance parameters for the plurality of tasks, the plurality of tasks having been supplied for performance by one or more task requesters and having each been performed by one or more of plurality of task performer users, the performance parameters for the plurality of tasks each corresponding to one or more aspects of the performance of the plurality of tasks, the automatic analyzing of the information being performed by the configured computing system;

automatically determining predicted information for one or more performance parameters for a future performance of one or more tasks that are distinct from the plurality of tasks and that have not yet been performed, the automatic determining being performed by the configured computing system based at least in part on one or more attributes of the one or more tasks and on the identified relationships; and

providing the determined predicted information for the one or more performance parameters so as to enable enhanced control of future performance of one or more tasks.

19. The non-transitory computer-readable medium of claim 18 wherein the computing system is at least part of a group of one or more computing systems that provide a task exchange server to facilitate performance by multiple human task performer users of multiple tasks supplied by multiple task requesters, wherein the attributes corresponding to the plurality of tasks reflect information about the plurality of tasks that is available before the performance of the tasks, wherein the automatic determining of the predicted information for the one or more performance parameters includes dynamically predicting likely values for the one or more performance parameters that will occur during the future performance in response to an indication of the one or more tasks that is received from a task requester, wherein the future performance of the one or more tasks represents performance by a subset of one or more of the multiple human task performer users that is unknown at a time of the dynamic predicting, and wherein the providing of the determined predicted information includes providing the determined predicted information to the task requester that indicated the one or more tasks so that the task requester may control the future performance of the one or more tasks by altering at least one of the one or more attributes of the one or more tasks before the future performance.

20. The non-transitory computer-readable medium of claim 18 wherein the computer-readable medium is a memory of the configured computing system.

21. The non-transitory computer-readable medium of claim 18 wherein the contents are instructions that when executed program the configured computing system to perform the method.

22. A computing system configured to automatically facilitate performance of tasks by task performer users based on predicted task performance information, comprising:

one or more memories; and

a task exchange server system configured to facilitate performance of tasks by users, by;

automatically analyzing information about prior performance of a plurality of tasks in order to identify relationships between attributes corresponding to the plurality of tasks and aspects corresponding to the performance of the plurality of tasks, the plurality of tasks being supplied for performance by one or more task requesters and being performed by a plurality of task performer users, the attributes reflecting information about the plurality of tasks that is available before the performance of the tasks;

automatically determining information about a predicted future performance of one or more tasks, the automatic determining being based at least in part on one or more attributes of the one or more tasks and on the identified relationships; and

providing the determined predicted future performance information so as to enable enhanced control of actual future performance of the one or more tasks.

23. The computing system of claim 22 wherein the task exchange server system facilitates performance by multiple human task performer users of tasks that are each supplied by one of multiple task requesters, wherein the automatic determining of the information about a predicted future performance of one or more tasks includes dynamically predicting likely performance parameter values for the actual future performance of the one or more tasks in response to an indication from a task requester of the one or more tasks, the actual future performance being estimated to be performed by a subset of one or more of the multiple human task performer users that is unknown at a time of the dynamic predicting, and wherein the providing of the determined predicted future performance information includes providing the determined predicted performance parameter values to the task requester that indicated the one or more tasks so that the task requester may control the actual future performance of the one or more tasks by altering at least one of the one or more attributes of the one or more tasks before the actual future performance.

24. The computing system of claim 22 wherein the task exchange server system includes software instructions for execution by the computing system.