US 7,197,459 B1Granted

Hybrid machine/human computing arrangement

AssigneeAmazon Technologies, Inc.

InventorsVenky Harinarayan, Anand Rajaraman, Anand Ranganathan

Life of this patent

Estimating the term…

Loading patent…

Abstract

A hybrid machine/human computing arrangement which advantageously involves humans to assist a computer to solve particular tasks, allowing the computer to solve the tasks more efficiently. In one embodiment, a computer system decomposes a task, such as, for example, image or speech comparison, into subtasks for human performance, and requests the performances. The computer system programmatically conveys the request to a central coordinating server of the hybrid machine/human computing arrangement, which in turn dispatches the subtasks to personal computers operated by the humans. The humans perform the subtasks and provide the results back to the server, which receives the responses, and generates a result for the task based at least in part on the results of the human performances.

Claims85

Dependent claims

A method for a computer system to use human assistance in performing tasks, the method comprising:

automatically and under control of a first computer system, causing a task to be performed by,
identifying a first and a second subtask of the task, wherein the first subtask involves speech recognition;
retrieving information about past quality of results of one or more humans when previously performing subtasks other than the first subtask;
using the retrieved past quality information to facilitate performance of the first subtask by one or more humans, the facilitating of the performance of the first subtask including,
identifying one or more required capabilities for performance of the first subtask;
dispatching the first subtask to a remote second computer system of a first human for performance by the first human, the first human identified as being one of one or more humans who have capabilities that satisfy the required capabilities for the first subtask, the retrieved past quality information including past quality information for the first human when previously performing multiple subtasks other than the first subtask, the past quality information for the first human being at least part of the capabilities of the first human that satisfy the required capabilities for the first subtask;
and
receiving a first result from the first human via the second computer system, the first result generated by performance of the first subtask by the first human;
providing payment to the first human for the performance of the first subtask;
and
generating a result for the task based at least in part on the first result.

8 more elements

An apparatus comprising:

a storage medium having stored therein a plurality of programming instructions that are machine executable, wherein when executed, said instructions operate to:
receive indications of a first and a second subtask of a task, the first subtask associated with one or more capabilities to be used in performance of the first subtask and having associated compensation for performance of the first subtask, wherein said task is one of speech comparison, speech recognition, conversion of speech into text, and comparison of music samples;
dispatch information about the first subtask to a remote computer system of a first human for performance by said first human of the first subtask, the first human identified as having one or more capabilities that satisfy the associated capabilities for the first subtask, the dispatched information including an indication of the associated compensation for performance of the first subtask;
receive a first result from the remote computer system based on performance of the first subtask by said first human;
retrieve information about past accuracy of said first human in previously performing subtasks;
and
generate a result for said task based at least in part on said first result and at least in part on the past accuracy information for said first human;
and
a processor coupled to said storage medium to execute said instructions.

6 more elements

A storage medium having stored therein a plurality of programming instructions that are executable to cause a first computing system to perform a method comprising:

receiving indications of a first and a second subtask of a task, the first subtask associated with one or more capabilities to be used in performance of the first subtask and having associated compensation for performance of the first subtask, wherein said task is one of speech comparison, speech recognition, conversion of speech into text, and comparison of music samples;
dispatching information about the first subtask to a remote computer system of a first human for performance by said first human of the first subtask, the first human identified as having one or more capabilities that satisfy the associated capabilities for the first subtask;
receiving a first result from the remote computer system based on performance of the first subtask by said first human;
retrieving information about past accuracy of said first human in previously performing subtasks;
and
generating a result for said task based at least in part on said first result and at least in part on the past accuracy information for said first human.

3 more elements

Dependencies are inferred from the claim text.

Facts

Application: 09/976,717Filed 12 Oct 2001 · us provisional application
Priority: 19 Mar 2001
Granted: 27 Mar 2007
Term ends: Not estimatedThe source records don’t support an estimate.

Classification5 codes

G10L 15/22
Speech recognition: procedures used during a speech recognition process, e.g. man-machine dialogue
G06Q 10/063112
Scheduling, planning or task assignment for a person or group: skill-based matching of a person or a group to a task
G06Q 10/10
Administration; Management: office automation; Time management
G10L
Speech analysis techniques or speech synthesis