Amazon Technologies, Inc.

Unwired Planet, LLC

Lenovo (beijing) Limited

Openwave Systems Inc.

Thomas Jay Hoover

Alexander Michael Spinelli

Luis Antonio Diniz Fernandes De Morais Sarmento

Lisa Jane Hinegardner

Ethan Lowry

Joseph Edward Heitzeberg

Jason Rukman

Patrick Greer Franklin

Edward Thomas Banti

Adam Callahan Sanders

Stefan Schroedl

Nick Ciubotariu

Srinivas Palla

Albert J Kennis

Lakshya Bhagat

Wei Zhang

Neil Christopher Fritz

Joseph Bradley

Aravindhan Rathakrishnan

Andrei Dorin Zaharia

Michael Albert Galassi

William Kevan Bennett Kennedy

Anupam Kumar

Nathan Andrew Sterken

Andrew James Will

Techniques for implementing a virtual assistant using directed acyclic graphs (DAGs) are described. When a trigger event occurs (e.g., the receipt of a natural language input, the receipt of sensor data, the detection of a performed gesture, etc.), an endpoint of processing of the trigger event is determined. A DAG is then generated that represents processing of the trigger event. Generate of the DAG starts by building a portion of the DAG for the endpoint processing component, and working backward based on data dependencies. Within the DAG, one or more components that output data, but do not receive data output by another component, may be determined. Thereafter, processing of the trigger event may be commences by initiating processing of the determined one or more components.

Processing orchestration for systems including machine-learned components

Devices and techniques are generally described for a mutation architecture for a contextual data aggregator. In some examples, a context aggregator system may determine a first operation associated with a first directive. In at least some examples, the first operation may be operable to store contextual data. In some examples, the context aggregator system may receive a first command including the first directive. In various examples, the context aggregator system may generate a first extension supporting the first operation based at least in part on the first directive. In some examples, the context aggregator system may receive a second command comprising the first extension and first contextual data. In at least some examples, the context aggregator system may store the first contextual data in a database based at least in part on the first extension.

Mutation architecture for contextual data aggregator

Systems including a universal context aggregator configured to pre-fetch context information that may be used to perform various processes with respect to a user input are described. The aggregator may have access to data representing what context information components of the system routinely request in various situations. When a particular situation is present, prior to being queried, the aggregator may pre-fetch context information that the aggregator is likely to be queried for.

Obtaining context data

Disclosed are various embodiments that involve selecting supplemental content. A compatibility rule is determined to require supplemental content from a first content provider to be a predetermined minimum distance from supplemental content from a second content provider. A layout is generated to include a first slot and a second slot. A first content item from the first content provider is assigned to the first slot, and a second content item from the second content provider is assigned to the second slot. A network page is generated using the layout.

Selecting supplemental content for inclusion in a network page

Disclosed are various embodiments for selecting page layouts based upon an outcome prediction associated with a request for a content page. Session variables associated with a request can be extracted. A score can be calculated based upon the session variables. The score can be generated by a machine learning engine that is trained using archived session data. A page layout can be selected based upon the generated score and a respective content page generated.

Generating a page layout based upon analysis of session variables with respect to a client device

Devices and techniques are generally described for shared execution caching of contextual data. A first request to retrieve a first value from a first field of a database may be received. First entity identifier data associated with the first request may be determined. A second request to retrieve the first value from the first field may be received, the second request being associated with second entity identifier data. A determination may be made that a first host device is associated with the first entity identifier data and the second entity identifier data. In some examples, a first executable instruction effective to retrieve the first value from the first field of the database when executed by the first host device may be generated. In some examples, the first value may be retrieved from the first field of the database and sent to the first speech processing component.

Speech processing using contextual data

Techniques for providing category predictions may be provided. For example, a process may attempt to improve a user experience when the user provides a search query. The process can predict the category associated with the search query, even when the category is not a keyword in the search query. Once the category is determined, data may be provided for the particular category, including data that enables an adjustment of a user experience. For example, when the category is apparel, the user experience may include an image-heavy layout and, when the category is books, the user experience may provide more text.

Category predictions with browse node probabilities

Method medium and system for category prediction for a changed shopping mission

A method and system for analyzing user behavior as users search for items within an electronic marketplace is provided. A query is submitted by a user of the electronic marketplace, the query is processed to identify a series of actions or behaviors performed by the user in relation to the query and fingerprint information for the query is determined based at least in part on analyzing the actions. In one embodiment, the electronic marketplace modifies a user experience for the user based on the fingerprint information. An interactive network region comprising search results related to a query issued by a user is generated, based on fingerprint information. Various categories of items related to a query issued by the user are identified based on fingerprint information and the search results are organized based on categories.

Category ranking based on query fingerprints

Embodiments of methods, systems, and storage medium associated with processing a search request for items are disclosed herein. In one instance, the method may include determining that a search request includes an indication of intent to obtain information about items that are related to other items, obtaining information about a first set of items from a first data repository, including the obtained information as input in the search request, and obtaining, in response to the search request, information about a second set of items related to the items of the first set from a second data repository. The second data repository may include information about items, including relationship of the second set of items with the first set of items. The item relationship may be determined based on collaborative filtering of information associated with the first or second items. Other embodiments may be described and/or claimed.

Refining user search for items related to other items

A method and system for analyzing user behavior as users search for items within an electronic marketplace is provided. A query is submitted by a user of the electronic marketplace, the query is processed to identify a series of actions or behaviors performed by the user in relation to the query and fingerprint information for the query is determined based at least in part on analyzing the actions. A classification for a query is determined based on the fingerprint information. In one embodiment, an electronic marketplace receives a query from a user, accesses a query classification database having fingerprint information, determines a fingerprint for the query based on the fingerprint information and dynamically modifies a user experience for the user based at least in part on the fingerprint information.

Generation and classification of query fingerprints

Disclosed are various embodiments that involve selecting supplemental content to be included with search results in a network page. At least one search result is generated by executing a search based at least in part on relevance to a search query. Supplemental content is selected based at least in part upon relevance to the search query without regard to the at least one search result. A subset of the supplemental content is selected in response to determining that the subset is permitted to be presented together in a search results page as defined by at least one compatibility rule. The search results page is generated to include the subset of the supplemental content and the search results.

Selecting supplemental content for inclusion in a search results page

Category predictions for user behavior

Category predictions identifying a search frequency

A method and system for analyzing user behavior as users search for items within an electronic marketplace is provided. A query is submitted by a user of the electronic marketplace, the query is processed to identify a series of actions or behaviors performed by the user in relation to the query and fingerprint information for the query is determined based at least in part on analyzing the actions. In one embodiment, the electronic marketplace modifies a user experience for the user based on the fingerprint information. Search results presented to the user are modified by arranging items of the search results in accordance with one or more layouts and views based on query fingerprint information. Various categories of items related to a query issued by the user are identified based on fingerprint information and the search results are organized based on categories.

Growing community of inventors

Seattle, WA, United States of America

Thomas Jay Hoover