Wetpaint.com, Inc.

Microsoft Technology Licensing, LLC

Hungryroot, Inc.

Alex David Weinstein

Dmitry Frenkel

Andrew Rothbart

Bo Steven Finneman

Daniel Joseph Giambalvo

Jennifer Marie Lin

Werner Koepf

Christopher Edgar Newell

Sean Cappello

Charles Cummins

Jeffrey Jon Weir

Timothy R Sullivan

Antoine Joseph Atallah

Daniel Tomko

Katie Messerly

Da Kong

Luke Joseph Vaccaro

Dmitry Yevgenyevich Ryabkov

Benjamin Pierce McKean

Rachel Lynn Donze

Jiaming Zhang

Alex Ingerman

Systems and methods to predict and optimize automated recipe selection and item delivery for a customer. An item-selection server selects a plurality of predicted recipes based on customer preferences and order history. Hard and soft constraints are obtained to perform the optimization of the predicted recipes. The hard constraints define parameters that cannot be violated by a recipe or set of recipes, and soft constraints define parameters that expresses a tradeoff value associated with a recipe or set of recipes. Optimization includes selecting multiple sets of predicted recipes based on the hard constraints and scoring each recipe set based on the soft constraints, item inventory, and customer preferences. A specific set of recipes is selected for the order based on the scores, and the filling of the order is initiated for the customer with items associated with the selected set of recipes.

System, method, and non-transitory computer-readable medium for artificial intelligence pre-filling grocery cart

A scene comprising a set of visual elements may allow a user to perform 'zoom' operations in order to navigate the depth of the scene. The 'zoom' semantic is often applied to simulate optical visual depth, wherein the visual elements are presented with different visual dimensions and visual resolution to simulate physical proximity or distance. However, the &#x201c;zoom&#x201d; semantic may be alternatively applied to other aspects of the visual elements of a scene, such as a user selection of a zoomed-in visual element, a &#x201c;drill-down&#x201d; operation on a data set, or navigation through a portal in a first data set to view a second data set. These alternative &#x201c;zoom&#x201d; semantics may be achieved by presenting the effects of a &#x201c;zoom&#x201d; operation within the scene on the visual presentation of the visual element in a manner other than an adjustment of the visual dimensions and resolution of the visual element.

Alternative semantics for zoom operations in a zoomable scene

Embodiments are directed towards determining an overall success of a plurality of features and recommending a subset of those features to a user for generating new content. For each feature that is associated with content, an overall success may be determined based on a success of each of a plurality of content that the feature is associated. A subset of the features may be displayed to the user&#x2014;from which the user can select features to utilize for generating the new content. Based on the features selected by the user, the displayed features may be dynamically updated to include one or more recommended features. A feature may be recommended to the user based on a combination of the overall success of each selected feature and the overall success of each recommended feature. The selected features may be utilized to create the new content.

Optimizing content production based upon distribution channel metrics

Embodiments are directed towards personalizing content to be provided to a user. A recommendation score may be determined for a piece of content for a user. The recommendation score may be based on a combination of an intrinsic value and/or a social value of the content to the user. The social value may be calculated based on a combination of an individual social value for each of the user's friends, which may be determined based on the combination of a social weight, an interest probability, and a recommendation score for the friend. Online personalization of content for a user may provide the user with the tools to be a valuable, appreciated member of the user's social network. By employing embodiments, as described herein, content may be determined and provided to a user, which may help the user gain the attention of their social circles.

Personalization based upon social value in online media

Embodiments are directed towards modifying when to publish content to a channel for distribution based on monitored actions associated with previously published content. A plurality of publishing slots may be determined from a plurality of time slots. Each publishing slot may indicate a time to publish content to the channel. Content may be published to the channel at each of the plurality of publishing slots. User actions associated with the published content may be monitored. The plurality of publishing slots may be ranked based on these monitored actions. At least one of the plurality of publishing slots may be modified to a different time slot based on the rankings of the publishing slots. The publication of content at a publishing slot may be delayed based on a success of other content published at a previous publishing slot being above a threshold.

Portfolio optimization for media merchandizing

Embodiments are directed towards dynamically managing contextual recommendations that can be employed with content when provided to a channel. A test for a variable having at least two outcomes may be determined to be employed with selected content. Each outcome may correspond to a different contextual recommendation. An outcome to employ with the content may be randomly determined/selected such that each of the outcomes of the test is selected a substantially similar number of times. The content may be modified and/or provided to the channel based on contextual recommendations that correspond to the determined outcome. Actions associated with a plurality of content may be monitored and/or analyzed to determine metrics for each outcome of a test. A comparison of the metrics for each outcome may be utilized to determine a preferred outcome for the test.

Dynamic playbook: experimentation platform for social networks

Embodiments are disclosed that relate to exposing metadata relationships through filter interplay. One embodiment provides a method comprising filtering a collection of items according to a first and a second metadata category to form an original subset of items that comprise metadata values for each of the categories. The method further comprises displaying a first and a second interactive filter view illustrating the original subset arranged according to their respective metadata values. The method further comprises receiving user input modifying an original range of metadata values of the first metadata category to form a modified range, determining a modified subset of items that comprise metadata values of the first metadata category falling within the modified range and that also comprise metadata values of the second metadata category, and modifying the second interactive filter view to illustrate the modified subset.

Exposing metadata relationships through filter interplay

Architecture for comparing images by building an initial map from the average color and an inserted blackened area. Accordingly, a map can be built that is more information-rich and smaller, thereby making the system more efficient. The architecture employs a Kohonen neural network (or self-organizing map (SOM)) by guiding the learning of the SOM using characteristics of the images such as average color and a central area. A strong component of the average color of the image and the central area at the approximate center of the image are added to the uninitialized SOM, which allows related colors to converge toward the central area of the image. When input, the SOM organizes the color content of the image on a map, which can be used to compare the image with other images.

Image analysis through neural network using image average color

Technologies are described herein for suggesting long-tail tags. A first group of tags and a second group of tags are identified from a plurality of tags. The first group of tags includes frequently-assigned tags having a higher frequency of being assigned to an asset. The second group of tags includes long-tail tags having a lower frequency of being assigned to the asset than the frequently-assigned tags. The frequently-assigned tags and a sample of the long-tail tags are suggested to a user upon receiving a request from the user to tag the asset.

Growing community of inventors

Bellevue, WA, United States of America

Alex David Weinstein