Abstract: Disclosed are various embodiments for inferring brand similarities using graph neural networks and selection prediction. In one embodiment, a brand-to-brand graph is generated indicating similarities between a set of brands according to at least one of: click-through data or conversion data. Using a first graph neural network (GNN) tower, the brand-to-brand graph is analyzed to determine brand similarities among a first brand identified from a search query and a first set of other brands. Using a second GNN tower, the brand-to-brand graph is analyzed to determine brand similarities among a second brand and a second set of other brands. A level of similarity between the first brand and the second brand is determined based at least in part on an output of the first GNN tower and an output of the second GNN tower.

Abstract: The present disclosure provides a method and apparatus for building a database for retrieval. An implementation of the method comprises: acquiring a data set, and dividing the data set into a first data set and a second data set; clustering the data in the first data set, to obtain at least one first-level cluster center; clustering the data in the first data set based on the first-level cluster center, to obtain corresponding at least one second-level cluster center; obtaining a codebook corresponding to the first data set based on residuals between the data in the first data set and the first-level cluster center and residuals between the data in the first data set and the second-level cluster center; and training the second data set based on the codebook corresponding to the first data set, to obtain a codebook corresponding to the data set.

Abstract: A computer-implemented method for storage of digital information data via at least one processing unit (110) operatively coupled to at least one database is proposed. In at least one embodiment, the method comprises: providing at least one portion of the digital information data; performing at least one syntactic and/or semantic search in the at least one database based upon the portion of the digital information data; providing one or more meta-data strings in response to the at least one syntactic and/or semantic search; receiving at least one relevant meta-data string; and storing the portion of digital information data and the at least one relevant meta-data string. In at least one embodiment, the at least one relevant meta-data string is usable for a future syntactic and/or semantic search.

Abstract: A computer-implemented method for template generation may include: receiving a first plurality of variables, each of the first plurality of variables specifying a feature to be implemented in a new template for document generation; for each of a plurality of existing templates for document generation, determining a degree of similarity between the first plurality of variables and a plurality of variables included in the respective existing template; upon determining that none of the degrees of similarity respectively determined for the plurality of existing templates satisfies a similarity threshold, determining that a combination of variables from two or more of the plurality of existing templates has a degree of similarity with the first plurality of variables satisfying the similarity threshold; and generating the new template based on the two or more of the plurality of existing templates.

Abstract: A method may include receiving a definition of a table user-defined function (TUDF) in a graph query language. The table user-defined function may be created based on the definition. For example, the creation of the table user-defined function may include checking and compiling the definition to generate executable code associated with the table user-defined function. Upon receiving a query including a relational query language statement invoking the table user-defined function, such as a structured query language select statement, the query may be executed on at least a portion of a graph data stored in a database. The executing of the query may include calling the executable code to execute the table user-defined function included in the relational query language statement. Related systems and computer program products are also provided.

Abstract: The present disclosure is directed toward systems, methods, and non-transitory computer readable media for generating and suggesting content collections for user accounts of a content management system using combinations of content-based features such as textual signals and visual signals. In some embodiments, the disclosed systems select a seed content item from among a plurality of content items associated with a user account within a content management system. From the seed content item, the disclosed systems can determine one or more germane topics and can cluster additional content items in relation to the germane topic(s). In addition, the disclosed systems can select one or more content items from a content cluster to provide as a suggested content collection.

Abstract: A Bayesian test of demographic parity for learning to rank may be applied to determine ranking modifications. A fairness control system receiving a ranking of items may apply Bayes factors to determine a likelihood of bias for the ranking. These Bayes factors may include a factor for determining bias in each item and a factor for determining bias in the ranking of the items. An indicator of bias may be generated using the applied Bayes factors and the fairness control system may modify the ranking if the determines likelihood of bias satisfies modification criteria for the ranking.

Abstract: A device with an identifier is described. The identifier includes a complete link for access to a digital playback platform of a multimedia file. The identifier is a readable identifier including a near-field communication (NFC) transceiver module, a QR code, or a barcode. Multimedia file configuration and multimedia file playback methods and systems on the digital playback platform are also described.

Abstract: A method and system for receiving data relating to one or more activities performed by a user on a document within a specific time period, the one or more activities being performed by using an application, analyzing the data to identify a category of user activity based at least on the type of activity performed on the document, and transmitting a signal to a device for storage in association with the document, the signal including the identified category.

Abstract: Disclosed embodiments may include a system for providing a nearest neighbors classification pipeline with automated dimensionality reduction. The system may receive a dataset. The system may determine whether the dataset has a first dimensionality that exceeds a predetermined threshold. When dataset has a dimensionality that exceeds a predetermined threshold, the system may prompt a user to input an explained variance threshold ratio. The system may receive the explained variance threshold ratio. The system may iteratively perform a binary search on the dataset to determine a reduced dimensionality having a total explained variance ratio closest to but not less than the explained variance threshold ratio. The system may reduce the dataset to the reduced dimensionality to generate a reduced dataset. The system may train a machine learning model using the reduced dataset.

Abstract: A data classification apparatus includes a data transformation unit that generates a feature vector by using classification target data, a classification estimation process observation unit that acquires, from a classification estimation unit that estimates classification of the classification target data and including a plurality of weak classifiers, observation information in a classification process based on the feature vector, and generates a classification estimation process feature vector based on the observation information, and an error determination unit that determines, in accordance with an input of the classification estimation process feature vector generated by the classification estimation process observation unit and a classification result output from the classification estimation unit to which the feature vector is input, whether the classification result is correct.

Abstract: Datasets can be synchronized across cloud-based computer environments using computer systems. A data map can be generated for source datasets for a plurality of computing environments, respectively, and an activation function can be determined based on the data map, for simultaneously activating a new dataset of a source system from one of the plurality of computing environments. A generation ID (identification) can be shared from the source system with one or more computing environments of the plurality of the computing environments, in response to receiving the activation function. The new dataset of the source system can be replicated across the one or more computing environments. A data synchronization trigger can be activated, using a synchronization coordinator module of the computer, and the data synchronization can be completed across the plurality of computing environments.

Abstract: Embodiments of the present disclosure relate to a method, an electronic device, and a computer program product for retrieving service requests. The method includes determining a hash value of a service request based on the service request. The method further includes determining request pairs based on correlations between the hash value of the service request and a plurality of hash values of a plurality of historical service requests. The method further includes determining a semantic correlation between the service request and the historical service request in each of the plurality of request pairs. The method further includes determining, based on the determined semantic correlation between each request pair, a probability indicating that the service request and the historical service request in the request pair use the same solution. The method further includes determining a retrieved historical service request based on the probability.

Abstract: Examples of the present disclosure describe systems and methods for generating multiple semantic hypotheses for search query intent understanding. In aspects, a search query may be received by a query analysis component associated with a search system. The query analysis component may be used to evaluate the search query for ambiguity in the domain, intent, and/or slot(s) of the search query. A set of hypotheses representing for one or more combinations of the domain, intent, and/or slot(s) of the search query may be generated. The set of hypotheses may be scored and/or ranked. Based on the scores/ranks, one or more of the hypotheses in the set of hypotheses may be provided to a user and/or one or more processing components accessible to the search system.

Abstract: A system can include instructions for generating a machine-learning model, including accessing, from a memory device, training data including two or more first commands. The first commands can be manually generated in response to a utilization exceeding a preset threshold and two or more first conditions, including a date and time the utilization exceeded the preset threshold. The training data can also include second commands generated by the machine-learning model in response to the utilization exceeding the preset threshold. The system may train the machine-learning model using a machine-learning algorithm using the training data and the second commands. The system may receive an indication of a utilization exceeding a preset threshold and determine, using the machine-learning model, an action responsive to the utilization. A command can be outputted to execute the action responsive to the utilization.

Abstract: An autonomous compute storage device system includes a computing device and a storage device that is coupled to the computing device. The storage device performs a first background operation for a storage subsystem that is included in the storage device and copies first data from the storage subsystem to a memory subsystem that is accessible to the storage device as part of the performance of the first background operation. If the storage device determines that a first autonomous compute signature matches the first data that was copied to the memory subsystem, it executes a first autonomous compute application to perform first compute operations that are associated with the first data that was copied to the memory subsystem during the performance of the first background operation and generate at least one first compute operation result.

Abstract: Systems and methods are disclosed herein for machine-learned vehicle desking operations. A vehicle recommendation system receives a request to determine similarities between vehicles. The request can indicate an identifier of a user-specified vehicle associated with vehicle attribute values (e.g., white color, sedan body style, 2020 manufacturing year, etc.). A machine learning model can determine respective embeddings for the vehicle attribute values and the respective embeddings can be concatenated, where the concatenated embeddings represent the user-specified vehicle in one embedding. The system can determine similarity metrics of the concatenated embeddings against reference embeddings. For example, a cosine similarity value can be determined for the concatenated embedding of the user-specified vehicle and the respective reference embeddings. Each similarity metric can represent a measure of similarity between the user-specified vehicle and a given vehicle.

Abstract: Embodiments are described herein that involve a media playback system playing back a playback queue according to a continuous-playback mode. In one aspect, a playback device is provided that includes a network interface, one or more media processing components, a non-transitory computer-readable medium with program instructions stored thereon. The program instructions are executable by at least one processor to cause the media playback device to: add one or more items to a playback queue of the media playback device, where the one or more items include a continuous-playback item including at least one media item, play back the playback queue according to a continuous-playback mode, where playing back the playback queue according to the continuous-playback mode involves: detecting a continuous-playback trigger, and based on the detected continuous-playback trigger, adding at least one additional media item to the continuous-playback item.

Abstract: A method and system for providing more relevant search results and recommendation from a food database is disclosed. The method includes receiving a query, a first candidate food, and a second candidate food. The method includes generating vectors based on the query and food names of the first and second candidate foods using at least one embedding function of a machine learning model. The method includes determining nutrition content vectors from the nutritional data of the first and second candidate foods. The method includes generating a nutrition content vector based on the query using another embedding function of the machine learning model. The method includes determining which of the first and second candidate food is more relevant to the query based at least in part on the nutrition content vectors. The method includes providing search results or recommendation based on the determined relevance.

Abstract: A business-document presenting device includes processing circuitry configured to store text information displayed on a terminal screen during processing of a task in a past, acquire text information displayed on the terminal screen, and compare the acquired text information and the text information stored in the storage and extract different items of text information among tasks as keywords for searching for business documents.