A centralized system may collect and aggregate assessments from multiple websites. An aggregate score may be calculated for the resource that cumulatively considers assessments from a plurality of different websites from which assessments are received from users. Text descriptions associated with each of the assessments may be provided to a machine-learning system that uses a trained model to assign identifiers to the assessments as they are received. These identifiers may include common words or text that are descriptive of different facets of user experiences related to receiving and using the resource. After selecting one or more identifiers, assessments associated with that identifier may be included or excluded from the display. Additionally, the overall aggregate score for the resource may be recalculated by removing components of that score that are based on assessments with identifiers that have been selected for exclusion.

Methods, apparatus, and processor-readable storage media for implementing dynamic widgets for displaying status information are provided herein. An example computer-implemented method includes executing a software widget configured to display, on a user device, status information associated with a plurality of items of an online data source; obtaining the status information from an application server via a first application programming interface, wherein the application server maintains at least a portion of the status information in a local database using a second application programming interface associated with the online data source; and updating a graphical user interface of the software widget to display the status information on the user device.

An online system presents content to its users, in which the content includes links that launch a messaging application and initiate conversations via the application. The system receives information indicating that negative experiences occurred during the conversations and may use this information and attributes of entities participating in the conversations to train a model to predict a likelihood of an occurrence of a negative experience. Upon determining an opportunity to present a user with a link that launches the application and initiates a potential conversation with an entity via the application, the system applies the model to predict a likelihood of an occurrence of the negative experience by the user during the potential conversation based on the entity's attributes. Based on the predicted likelihood, the system adjusts a value associated with presenting the link and passes the adjusted value to a process that selects content for presentation to the user.

An online system presents content to its users, in which the content includes links that launch a messaging application and initiate conversations via the application. The system receives information indicating that negative experiences occurred during the conversations and may use this information and attributes of entities participating in the conversations to train a model to predict a likelihood of an occurrence of a negative experience. Upon determining an opportunity to present a user with a link that launches the application and initiates a potential conversation with an entity via the application, the system applies the model to predict a likelihood of an occurrence of the negative experience by the user during the potential conversation based on the entity's attributes. Based on the predicted likelihood, the system adjusts a value associated with presenting the link and passes the adjusted value to a process that selects content for presentation to the user.

Structured documents are processed using convolutional neural networks. For example, the processing can include receiving a rendered form of a structured document; mapping a grid of cells to the rendered form; assigning a respective numeric embedding to each cell in the grid, comprising, for each cell: identifying content in the structured document that corresponds to a portion of the rendered form that is mapped to the cell, mapping the identified content to a numeric embedding for the identified content, and assigning the numeric embedding for the identified content to the cell; generating a matrix representation of the structured document from the numeric embeddings assigned to the cells of the grids; and generating neural network features of the structured document by processing the matrix representation of the structured document through a subnetwork comprising one or more convolutional neural network layers.

Systems and methods for identifying content for an input query are presented. A mapping model is trained to map elements of an input query embedding vector for a received query into one or more elements of a destination embedding vector. In response to receiving an input query, an input query embedding vector is generated that projects into an input query embedding space. The input query embedding vector is processed by the mapping model to map the input query embedding vector into one or more elements of a destination embedding vector in a destination embedding space, resulting in a partial destination embedding vector. Items of a corpus of content are projected into the destination embedding space and the partial destination embedding vector is also projected into the destination embedding space. A similarity measure determines the most-similar items to the partial destination embedding vector and at least some of the most-similar items are returned in response to the input query.

Systems and methods for identifying content for an input query are presented. A mapping model is trained to map elements of an input query embedding vector for a received query into one or more elements of a destination embedding vector. In response to receiving an input query, an input query embedding vector is generated that projects into an input query embedding space. The input query embedding vector is processed by the mapping model to map the input query embedding vector into one or more elements of a destination embedding vector in a destination embedding space, resulting in a partial destination embedding vector. Items of a corpus of content are projected into the destination embedding space and the partial destination embedding vector is also projected into the destination embedding space. A similarity measure determines the most-similar items to the partial destination embedding vector and at least some of the most-similar items are returned in response to the input query.

Method and system to automatically verify the layout of a webpage on various screen sizes. The method includes retrieving screen size data associated with the webpage. The method further includes determining anchor points for the webpage based on the retrieved screen size data. The method also includes rendering the plurality of elements of the webpage based on the retrieved screen size data and the determined anchor points and identifying a respective location of each of the plurality of elements of the webpage to verify the layout of the plurality of elements of the webpage. If an error in the layout of the webpage is detected, a visual and/or audio alert is sent.

Techniques for tuning a machine learning algorithm using automatically determined optimal hyperparameters are described. An exemplary method includes receiving a request to determine a search space for at least one hyperparameter of a machine learning algorithm; determining, according to the request, optimal hyperparameter values from the search space for at least the one hyperparameter of the machine learning algorithm based on an evaluation of hyperparameters from the same machine learning algorithm on different datasets; and tuning the machine learning algorithm using the determined optimal hyperparameter values for the at least one hyperparameter of the machine learning algorithm to generate a machine learning model.

A method comprises receiving a first outbound request, from an internal user account of an internal platform, indicating a first action to be performed by a first third-party user account of a first third-party platform. In response to authenticating the first outbound request, the method further comprises sending an application programming interface (API) request to the first third-party platform to perform the first action on the first third-party platform on behalf of the internal user account. The method further comprises receiving a first inbound request, from the first third-party user account, indicating a second action to be performed on behalf of the internal user account on the internal platform. In response to authenticating the first inbound request, the method further comprises sending an internal request to the internal platform to perform the second action on the internal platform on behalf of the first third-party user account.