Abstract: Variable patch shape synthesis techniques are described. In one or more implementations, a plurality of patches are computed from one or more images, at least one of the plurality of patches having a different shape than another one of the plurality of patches. The shapes define an area to be considered for use in a patch synthesis technique. The patch synthesis technique is performed to edit an image using the computed plurality of patches having the different shapes.

Abstract: The invention is directed towards segmenting images based on natural language phrases. An image and an n-gram, including a sequence of tokens, are received. An encoding of image features and a sequence of token vectors are generated. A fully convolutional neural network identifies and encodes the image features. A word embedding model generates the token vectors. A recurrent neural network (RNN) iteratively updates a segmentation map based on combinations of the image feature encoding and the token vectors. The segmentation map identifies which pixels are included in an image region referenced by the n-gram. A segmented image is generated based on the segmentation map. The RNN may be a convolutional multimodal RNN. A separate RNN, such as a long short-term memory network, may iteratively update an encoding of semantic features based on the order of tokens. The first RNN may update the segmentation map based on the semantic feature encoding.

Abstract: Systems and methods provide for analyzing a group of online articles to identify relevant and popular topics. Text from each online is analyzed to identify topics relevant to each online article and to generate a relevance score for each topic and each online article. The topics are scored as a function of the relevance scores and visitor metrics for the online articles. The visitor metrics may include all visitors or only visitors within a particular visitor segment. The most relevant and popular topics are identified based on the scored topics. In some embodiments, the online articles are further analyzed to identify terms used in the online articles that are important to each topic. Further embodiments analyze the online articles to determine the lifespan of topics and terms, reflecting the popularity of topics and terms over time intervals.

Abstract: Systems and methods disclosed herein facilitate adding content to forms by providing field content suggestions using context determined based on form features. This involves providing a user interface for display, the user interface displaying a form and providing editing features usable to add content to fields of the form. The systems and method further involve categorizing the form or portion of the form into a context based on form features for non-text field characteristics or field-specific text characteristics of the form or portion of the form. A form field suggestion for a field is identified based on the context of the form or portion of the form containing the field and the field content suggestion for the field is provided for display in the user interface.

Abstract: Assets or the locations where assets may appear are monitored. If a new asset appears within a monitored location or if an existing asset is copied, modified, or moved, a decision is made as to whether to assign a new unique identifier or to assign a new version number to an existing unique identifier. The unique identifiers and version numbers are carried as metadata within the assets and optionally also stored separately within a data store.

Abstract: Systems and methods provide for efficiently obtaining biometric signatures for electronically signing digital documents. A digital document having a signature field is provided for display on a general computing device. An instruction to obtain a biometric signature from a mobile computing device is received on the general computing device. A remote signing request is sent to a remote server device. The request can include, among other things, an electronic contact address associated with the mobile computing device. Responsive to receiving the request, a Uniform Resource Identifier (URI) is generated, referencing a dynamic application for obtaining the biometric signature on the mobile computing device. The remote server device obtains the biometric signature via the dynamic application or an incoming electronic message from the mobile computing device via the electronic contact address.

Abstract: A buffer model in an HTTP streaming client may include receiving a first content fragment of a first content stream in response to a first HTTP request. It may also include receiving a second content fragment of a second content stream in response to a second HTTP request. The buffer model may further include storing the first and second content fragments in first and second buffers of a plurality of configurable buffers. The first and second content fragments may be multiplexed into a third buffer of the plurality of buffers. The multiplexed first and second content fragments may be stored in a fourth buffer of the plurality of buffers for playback. The buffer model may be implemented by an application. The buffers may be designed based on one or more constraints.

Abstract: Various embodiments calculate a score for each frame of a video segment based on various subject-related factors associated with a subject (e.g., face or other object) captured in a frame relative to corresponding factors of the subject in other frames of the video segment. A highest-scoring frame from the video segment can then be extracted based on a comparison of the score of each frame of the video segment with the score of each other frame of the video segment, and the extracted frame can be transcoded as an image for display via a display device. The score calculation, extraction, and transcoding actions are performed automatically and without user intervention, which improves previous approaches that use a primarily manual, tedious, and time consuming approach.

Abstract: Timeline creation of electronic document creation states is described. In one or more implementations, document creation states are created responsive to performing editing operations on a document. Document creation state snapshots are then generated for respective document creation states responsive to the document creation states meeting at least one criterion. The at least one criterion may be based on an amount of change in the document, completion of an editing operation, specific alterations to the document, a user request to create a document creation state snapshot, and so on. The resulting document creation state snapshots along with visual representations of metadata associated with the respective document creation states are compiled into a single timeline in a chronological order of when the respective document creation states occurred.

Abstract: Responsive grid layouts for graphic design are described. In one example, a system of a digital medium environment is configured to generate responsive grid layouts for graphic design. The system includes a layout module implemented at least partially in hardware of the computing device to generate a plurality of grid layouts each having at least one cell configured to be mapped to at least one design element of digital content. The system also includes a scoring module implemented at least partially in hardware of a computing device to assign scores to the plurality of grid layouts based on aesthetic criteria and an output module implemented at least partially in hardware of a computing device to output at least one grid layout of the plurality of grid layouts for recommendation to a user based on the scores.

Abstract: Metric forecasting techniques and systems in a digital medium environment are described that leverage similarity of elements, one to another, in order to generate a forecast value for a metric for a particular element. In one example, training data is received that describes a time series of values of the metric for a plurality of elements. The model is trained to generate the forecast value of the metric, the training using machine learning of a neural network based on the training data. The training includes generating dimensional-transformation data configured to transform the training data into a simplified representation to determine similarity of the plurality of elements, one to another, with respect to the metric over the time series. The training also includes generating model parameters of the neural network based on the simplified representation to generate the forecast value of the metric.

Abstract: The present disclosure includes methods and systems for completing fields in an electronic document utilizing drawing input. In particular, upon receiving drawing input in an electronic document with a plurality of fields, one or more embodiments of the disclosed systems and methods automatically identify an appropriate field corresponding to the drawing input. For instance, the disclosed systems and methods can assign drawing input to an appropriate field based on a number of drawing units located within the boundary of each field in the plurality of fields. Moreover, upon identifying the appropriate field corresponding to the drawing input, the disclosed systems and methods can automatically modify the drawing input to fit within a boundary of the identified field.

Abstract: Systems and methods are disclosed for planning, executing, reviewing, and reporting the results of an advertising campaign to be run on TV. A demand-side platform receives ad slot opportunities from TV programming sources, and analyzes the ad slots to produce a prioritized list of placement opportunities for the advertising campaign to be presented to advertiser/clients. Each ad slot is analyzed with respect to past viewership data and with respect to desired targeting characteristics that may include conventional age and gender targeting, or additionally strategic targeting characteristics. Scores are established for each ad slot with respect to numbers of projected on-target impressions and/or a cost for projected on-target impressions. The scores are sorted to produce the prioritized list. Projected results can be viewed with respect to any or all of network, day, and daypart. After a campaign has completed, viewership data representing actual results is acquired, processed, and reported.

Abstract: Techniques are disclosed for auto-generating a preview of a web conference based on participant interactions during the web conference. A web conference is recorded and includes a shared presentation and participant interactions. Key frames of the shared presentation are identified. During the web conference, differing amounts of interaction between the participants occurred during presentation of the respective key frames. The technique selects a subset of the key frames to include in the preview based on the differing amounts of interaction between the participants that occurred during presentation of the key frames and generates the preview using the subset of key frames. An example preview displays a sequence of key frames in which each key frame is displayed for a few seconds.

Abstract: Digital image defect identification and correction techniques are described. In one example, a digital medium environment is configured to identify and correct a digital image defect through identification of a defect in a digital image using machine learning. The identification includes generating a plurality of defect type scores using a plurality of defect type identification models, as part of machine learning, for a plurality of different defect types and determining the digital image includes the defect based on the generated plurality of defect type scores. A correction is generated for the identified defect and the digital image is output as included the generated correction.

Abstract: A tool is provided via a user interface for a digital media application that supports digital illustrations. The tool combines operations to create different types of segments for a drawing shape and covert between types of segments. The tool is configured to analyze the drawing to recognize segments that are straight, arc, or curved portions of the drawing path. For segments recognized as curved, the segments are represented as Bezier curve segments. For segments recognized as straight, the segments are represented as line segments. Additionally, line segments are associated with handle elements operable to convert the line segments to regular arc segments. Responsive to manipulation of a handle element for a particular line segment, the tool computes a corresponding regular arc and converts the line segment into a regular arc segment.

Abstract: A computer implemented method and system identifies correct structured reading-order sequence of text segments that are extracted from a file structured in a portable document format. A probabilistic language model is generated from a large text corpus to comprise observed word sequence patterns for a given language. The language model measures whether splicing together a first text segment with another continuation text segment results in a phrase that is more likely than a phrase resulting from splicing together the first text segment with other continuation text segments. Sets of text segments are provided to the probabilistic model, where the sets of text segments comprise a first set including the first text segment and a first continuation text segment. A second set includes the first text segment and a second continuation text segment. A score is obtained for each set of text segments. The score is indicative of a likelihood of the set providing a correct structured reading-order sequence.

Abstract: Streaming digital content synchronization techniques are described. A response is received to a request to stream the digital content. The response includes a time at which the digital content was last modified (e.g., a last-modified header) and a time at which the response was generated, e.g., a date header. An age is calculated by subtracting the time at which the digital content was last modified, e.g., the last-modified header, from the time at which the response was generated, e.g., the date header. An age describing an amount of time the response spent in one or more caches, if available, is added as part of this age. The time is determined by subtracting the age from a predefined setback time and the stream of the digital content is rendered based at least in part on the determined time.

Abstract: User data and a plurality of micro-segment definitions are received. Each micro-segment definition in the plurality of micro-segment definitions corresponds to one or more offers in an offer provider campaign. Further, a each micro-segment definition from the plurality of micro-segment definitions is parsed into a plurality of parsed expression segments that indicate a plurality of micro-segment condition rules. The plurality of parsed expression segments are compiled into an executable object that indicates a plurality of instructions to determine if the user data matches the plurality of micro-segment definitions. Each micro-segment definition is processed to apply the plurality of micro-segment condition rules to the user data to determine a match of a user belonging to a micro-segment. Further, a score is assigned to indicate the strength of each match. In addition, each match is ranked according to the score for each match.

Abstract: Systems and methods are provided for authorizing third-party access to a specific service from a service provider. In an example embodiment, a server system identifies a shared service from multiple services provided by the server system. The shared service is specified by an authorizing entity. The server system provides a credential associated with the shared service and the authorizing entity. The server system receives a request to access the shared service from a requesting entity that is separate from the authorizing entity. The server system verifies that the request includes the credential and that the credential is associated with the shared service and the authorizing entity. The server system provides access to the shared service to the requesting entity based on verifying that the request includes the credential. The requesting entity is restricted to accessing the shared service identified by the credential as authorized by the authorizing entity.