Amazon Technologies, Inc.

Imdb. Com, Inc.

Nicole A Deflaux

James Cleveland Willeford

Adam D Bradley

Vidya V Iyer

Ranganath Atreya

Anand Chelian

Luis Felipe Cabrera

Adam Carlson

Joseph C Lee

Mark Chien

James Robert Blair

Jang Han Goo

Brian A Ecker

Mohamed Reza Hussein

Techniques are described for facilitating performance of tasks involving human review of media data (e.g., images, audio clips, video clips, etc.), such as to identify one or more targets of interest. The media data review may be performed for various reasons, such as to assist in locating one or more target objects in a large geographical area (e.g., one or more missing people or vehicles) based on human review of large quantities of images that each correspond to a small subset of the geographical area. When one or more of the reviewed media data pieces are identified as being relevant, information about those identified media data pieces may be used in various ways. Furthermore, in at least some situations, the media data review may be part of a charitable effort that uses human volunteers to review the media data without monetary payment or other compensation.

Managing performance of human review of media data

Supplemental content such as electronic advertisements may be previewed in situ. That is, supplemental content may be previewed in a context of a live and/or production environment along with the primary content available in that environment. In particular, electronic advertising that is still under development, that still requires testing and/or quality control, and/or electronic advertising that is unpublished may be incorporated into electronic pages by live and/or production electronic page generation computer systems. Furthermore, electronic advertising may be previewed in situ in a controlled and/or restricted manner.

Restricted in situ previews for electronic advertising

Techniques are described for facilitating access of users to tasks that are available to be performed, such as via a task exchange system that serves as an intermediary to facilitate performance by some task performer users of tasks supplied by other task requester users. The techniques may include automatically providing a particular group of task performer users with access to a supply of available tasks that are appropriate for those task performer users, such as based on an aggregate probability or other likelihood measure that the task performer users as a group will collectively accept to perform those tasks in a specified manner. More generally, the affinity and/or appropriateness of a group of users to perform one or more tasks may be based on various information related to the users, including past task performance behavior of those users, and/or qualifications or more general reputational aspects of those users.

Providing access to tasks that are available to be performed

Techniques are described for providing enhanced access to stored data and related functionality, such as data stored by a system across multiple alternative storage partitions. At least some of the stored data may be specific to particular users who interact with the system, such as for users to perform tasks supplied by other users. The enhanced data access techniques may include storing at least a subset of each user's associated data on a single storage partition, and then efficiently identifying the storage partition corresponding to a user so that requests from that user may be handled by a computing node associated with that corresponding storage partition. Furthermore, additional data access enhancement techniques may relate to initially identifying an appropriate storage partition for a user and/or for migrating existing user data to a new partition identified as being appropriate for a user.

Providing enhanced access to stored data

A scalable data store framework is provided to migrate data across partitions in a data store without compromising data availability for customers. The scalable data store framework includes one or more partitions configured to manage a set of 'entities' for a network-based service. When a new-partition event is detected, a target partition is identified for bifurcation and various additional partitions are added to the data store. After the target partition is replicated into the additional partitions, some entities associated with the target partition are migrated to the additional partitions in order to maintain good load balance in the data store.

Data store replication for entity based partition

Techniques are described for storing and accessing data on heterogeneous types of data repositories, such as by a distributed software system that uses multiple data repositories on multiple computing nodes, including to transfer groups of data between multiple heterogeneous types of data repositories. In some situations, the techniques may be used by a system that stores various types of data regarding users or other entities that are modeled by the system, such as to transfer a group of data that represents an entity. The transfer of data may be facilitated by use of an abstraction interface that provides a uniform interface for accessing the multiple data repository types, such as an abstraction interface provided by one or more storage management components that further provide functionality to translate data between various data formats used by the multiple data repository types, such as via use of a common data format.

Growing community of inventors

Seattle, WA, United States of America

Nicole A Deflaux