Abstract: In some implementations, the techniques described herein relate to a method including (i) receiving, by a processor, user input describing at least one parameter for a query image, (ii) generating, via a generative machine learning model executed by the processor, the query image based at least in part on the user input describing the at least one parameter for the query image, (iii) providing, by the processor, the query image as input to an image-based search algorithm, and (iv) returning a result received by the processor from the image-based search algorithm.

Abstract: The present teaching relates to method, system, medium, and implementations for identifying k nearest neighbors. A plurality of combined neighborhoods are received from a plurality of local join executors. Each combined neighborhood represents a neighborhood of a source data point and has one or more pairs of neighbors, each of which includes the source data point, a neighbor of the source point, and a distance in-between. A plurality of KNN lists corresponding to a plurality of source data points are obtained. Each KNN list includes K neighbors to a corresponding source data point, each of which is represented by an index of the neighbor and a distance between the source data point and the neighbor. The plurality of KNN lists are updated based on the plurality of combined neighborhoods to generate updated KNN lists.

Abstract: In some implementations, the techniques described herein relate to a method including: (i) identifying, by a processor, an electronic message addressed to an inbox of a user that comprises a confirmation of a transaction involving a platform, (ii) searching, by the processor, for an additional electronic message addressed to the inbox indicating an alteration applicable to the transaction, (iii) causing display, by the processor, of a prompt informing the user of the alteration applicable to the transaction, (iv) composing, by a large language model (LLM) executed by the processor, a potential electronic message to an operator of the platform requesting the alteration be retroactively applied to the transaction, and (v) in response to receiving user input regarding the potential electronic message, sending, by the processor, a subsequent electronic message to the operator of the platform requesting the alteration be retroactively applied to the transaction.

Abstract: A location-aware search assist capability identifies location-aware search query suggestions using location information associated with the location-aware search query suggestions. A user's search query input and location and a location associated with each location-aware search query suggestion candidates may be used to identify a set of search query suggestions for presentation to the user. Location-aware search query suggestion candidates may be ranked in accordance with a closeness of each one's location to the user's location. The ranking may be performed using a score, such as a popularity score associated with each search query suggestion candidate. The location-aware search query suggestion candidates having a location closer to the user's location may be promoted by adjusting each candidate's popularity score upward, and the search query suggestion candidates that are farther away from the user's location may be demoted by adjusting each such candidate's popularity score downward.

Abstract: The present teaching relates to searching. In one example, a search query is received from a person. A plurality of search results are retrieved based on the search query. An intent of the person is estimated with respect to at least some of the plurality of search results. The estimated intent is what the person intends to do with respect to the at least some of the plurality of search results. The plurality of search results are provided based on the estimated intent of the person.

Abstract: In some implementations, the techniques described herein relate to a method including: receiving a training data set, the training data set including data representing consumer interactions and actions taken by consumers after the consumer interactions; executing a training run using a predictive model, the predictive model including a plurality of trainable parameters; computing a loss of the training run using a loss function, the loss function including a strict convex function; optimizing the plurality of trainable parameters based on an output of the loss function; storing the trainable parameters as an inference model; and predicting a future ratio using the inference model.

Abstract: One or more computing devices, systems, and/or methods for implementing an automated model update pipeline are provided. User behavior data associated with content provided to users may be collected. An automatic model training is invoked to train a new model to output a set of model parameters based upon a configuration specifying a target audience, features extracted from user behavior data, and training model parameters. In response to determining that the new model will outperform a deployed model on a content serving platform, an automatic model updater is invoked to update the content serving platform with the new model and a ranking profile of the new model for serving content requests.

Abstract: The disclosed systems and methods provide a novel action prediction framework that performs personalized action prediction. According to an embodiment, the disclosed framework is able to dynamically predict which action (if any) a user might perform in response to receiving a given message. In some embodiments, for a given message, the action prediction framework can determine the probability that a user (e.g., sender, recipient) associated with the message may perform an action or set of action actions (e.g., open, forward, delete, reply, archive) related to the message. In some embodiments, the framework may be used to suggest a predicted action to the user. In some embodiments, a computing device may use the predicted actions to automatically perform the action. According to an embodiment, the action prediction framework includes a multi-label or multi-class model using a neural network.

Abstract: One or more computing devices, systems, and/or methods are provided. In an example, a first performance metric score may be determined based upon first content item text. A plurality of similarity scores associated with a plurality of sets of content item text may be determined. One or more sets of content item text may be selected from among the plurality of sets of content item text based upon the plurality of similarity scores and a plurality of performance metric scores associated with the plurality of sets of content item text. The plurality of performance metric scores may comprise one or more performance metric scores associated with the one or more sets of content item text. The one or more performance metric scores may be higher than the first performance metric score. One or more representations of the one or more sets of content item text may be displayed.

Abstract: Disclosed are systems and methods that provide a decision-intelligence (DI)-based, computerized framework for compiling and leveraging reliable sequence taggings for input queries related to executed searches. The disclosed framework can compile a trained computer model to fulfill partially labeled queries tagged by AI models as fully labeled queries. The disclosed framework can further leverage other AI models (e.g., deep neural networks, knowledge graphs, and the like), so that cross-checks can be performed between different models to guarantee high quality of labeled tokens. Thus, the framework can automatically generate and implement reliable training data to train a sequence tagging model for search query understanding. Thus, the search engine operating on such tagging model can provide improved results.

Abstract: One or more computing devices, systems, and/or methods for classifier validation are provided. A set of in-sample examples are partitioned into a reduced in-sample set and a remaining in-sample set. The reduced in-sample set is processed using a set of classifiers. A subset of classifiers are identified as having error counts, over the reduced in-sample set, below a threshold number of errors. A training procedure is executed to select a classifier having a minimum error rate over the set of in-sample examples. If the classifier is within the subset of classifiers, then an out-of-sample error bound is determined for the classifier.

Abstract: The present teaching relates to content recommendation. Content is selected from multiple pieces of content for recommending to a user based on a user embedding and contextual information. Performance of the user with respect to the recommended content is obtained. The user embedding for the user is adapted based on the performance information via a memory-dependency aware model previously trained to learn past interests of the user and intensities of such past interests. An updated user embedding is generated to represent current interests of the user via the adapted user embedding produced by the memory-dependency aware model according to the known intensities of the past interests of the user as well as user's interest exhibited in the recommended content.

Abstract: The present teaching relates to online user profiling. In one example, content associated with a first user of a social media network is obtained. From the content associated with the first user, a first link to a first piece of content is identified. A second user of the social media network associated with the first user is determined in the context of the social media network. From content associated with the second user of the social media network, a second link to a second piece of content is identified. The first and second pieces of content are retrieved based on the first and second links, respectively. User profile of the first user is generated based, at least in part, on the first and second pieces of content.

Abstract: Disclosed are systems and methods for improving interactions with and between computers in content searching, hosting and/or providing systems supported by or configured with devices, servers and/or platforms. The disclosed systems and methods provide a novel framework for providing users with electronic retrieval capabilities that are activated upon the users' determined locations respective to real-world locations associated with a message providing entity. The disclosed technology combines the previously separate systems of mail extraction, geo-fencing and content delivery (e.g., notification) into a single system that efficiently manages a user's inbox in order to provide the user with content the user has expressly indicated is of interest to that user.

Abstract: In an example, first entities are extracted from user profiles. Second entities are extracted from content information associated with content item. User-associated metrics associated with the first entities are determined based upon the user profiles and/or content events. First vector representations of the first entities and second vector representations of the second entities are processed to generate an attention distribution array. Each value of the attention distribution array represents, for a user interested in an entity of the first entities, a proportion of (i) entity-specific activity, of the user, related to an entity of the second entities relative to (ii) an entirety of activity of the user. An inferred activity distribution array is generated by applying the user-associated metrics to the attention distribution array. A filtered subset of activity distribution values is generated by pruning values from the inferred activity distribution array.

Abstract: Techniques for evaluating a user experience experiment designed to use one user experience variant selected from a number of user experience variants as a global-best user experience variant to be used across users relative to a machine model trained to use user data to identify a user-preferred user experience variant. Disclosed systems and methods provide techniques for optimizing user response. In one embodiment, a global-best user experience variant is evaluated by comparing an aggregate user response determined for the global-best user experience variant to an aggregate user response determined using user response predictions determined using the trained machine model, and using the outcome of the comparison to make a recommendation as to which one of the global-best user experience variant and the trained machine model to adopt for providing a user experience to users.

Abstract: The disclosed systems and methods provide a novel framework that provides mechanisms for performing cost-effective, accurate and scalable detection and recognition of fine-grained events. The framework functions by training high precision and high recall object/optical character recognition (OCR) models and aligning video frames to text commentaries of the videos (e.g., licensed play-by-play). The disclosed framework operates as a single algorithm that performs multimodal alignments between events/actions within videos and their prescribed text. Thus, the disclosed framework is able to scale to fine-grained action categories across different venues by delving into the key frames and key aspects of a video to identify particular actions performed by particular actors, thereby providing the novelty of fine-granted action detection and recognition.

Abstract: In accordance with the present disclosure, one or more computing devices and/or methods are provided. In an example, an emotional state of a user may be determined based upon one or more signals generated by one or more sensors and/or user activity of the user. A communication interface may be controlled based upon the emotional state.

Abstract: The present teaching relates to method and system for generating a stream of content items. A plurality of entities associated with a time-window are obtained, wherein each entity of the plurality of entities is associated with at least one content item. For each entity, a first parameter with respect to the time-window, a second parameter with respect to previous time-windows, and a trendiness score based on a function of the first parameter and the second parameter are respectively calculated. A graph based on one or more entity-pairs is generated, wherein each entity-pair of the one or more entity-pairs satisfies a first criterion. A stream of content items is generated based on the graph, wherein each content item in the stream of content items corresponds to at least one of the one or more entity pairs.

Abstract: Technologies for enhancing work scheduling of a big data framework. The technologies can include generating, in a database, configurable namespaces to be used by a work scheduling enhancement application to group together tasks of one or more big data clusters. The namespaces can be hierarchical. The technologies can also include linking in the database, by the application, related tasks with respective namespaces to categorize and group together the related tasks. The technologies can also include configuring, by the application, a display scheme for displaying error handling and root cause analysis of tasks of the one or more big data clusters. The technologies can also include generating or rendering, by the application, a GUI having a navigable hierarchal view for displaying the namespaces. The generation or rendering of the GUI can be based partially on the display scheme.