Abstract: Systems and methods for verifying identifies of parties to a video conference call are disclosed. An example method includes storing a plurality of first facial encodings each associated with one or more images of a registered agent, receiving, during the video conference call, a second facial encoding associated with one or more images of a person participating in the video conference call, determining that a specified registered agent is scheduled as a party to the video conference call, comparing the second facial encoding to a selected first facial encoding associated with the specified registered agent, and determining, based at least in part on the comparing, whether or not the person is the specified registered agent.

Abstract: A method for storing object state on hash chains. A block of a hash chain is located with an accessor function. The block includes a field value of a field of an object. Version data is located using the block. The version data identifies a block version of the block. The block version is compared to a function version of the accessor function. The field value from the block is returned when the block version matches the function version. A version matched accessor function is called to return the field value when the block version does not match the function version.

Abstract: Certain aspects of the present disclosure provide techniques for providing assistance to users by integrating social computing system with conversational user interface. In some cases, a user interacting with a virtual assistant of a conversational user interface provides input that the virtual assistant is not able identify a matching intent. As a result, the virtual assistant can leverage the social computing system to generate a new question based on the user input and post the question to the social computing system. Users of the social computing system can provide an answer, which the virtual assistant provides to the user in the conversational user interface. The social computing system can also generate a new intent for the virtual assistant to increase efficiency of the virtual assistant.

Abstract: A method of training machine learning models (MLMs). An issue vector is generated using an issue MLM to generate a first output including first embedded natural language issue statements. An action vector is generated using an action MLM to generate a second output including related embedded natural language action statements. The issue and action MLMs are of a same type. An inner product of the first and second output is calculated, forming a third output. The third output is processed according to a sigmoid gate process to predict whether a given issue statement and corresponding action statement relate to a same call, resulting in a fourth output. A loss function is calculated from the fourth output by comparing the fourth output to a known result. The issue MLM and the action MLM are modified using the loss function to obtain a trained issue MLM and a trained action MLM.

Abstract: Systems and methods for generating recommended offers are disclosed. An example method may be performed by one or more processors of a recommendation system and include correlating attributes of users with attributes of offers based on historical data associated with the users and offers, training a machine learning model to predict a user's interest in an offer based on the correlating, obtaining current user data, obtaining current offer data, providing the current user data and the current offer data to the trained machine learning model, generating, using the trained machine learning model, a predicted level of interest that the current user has in each respective current offer of the number of current offers, identifying, among the number of current offers, at least one current offer having a predicted level of interest for the current user greater than a value, and generating one or more recommended offers for the current user.

Abstract: Systems and methods that may be used to provide multitenant key derivation and management using a unique protocol in which key derivation may be executed between the server that holds the root key and a client that holds the derivation data and obtains an encryption key. In one or more embodiments, the derivation data may be hashed. The disclosed protocol ensures that the server does not get access to or learn anything about the client's derived key, while the client does not get access to or learn anything about the server's root key.

Abstract: Disclosed are techniques for classifying inquiries and providing recommended actions. These techniques identify or generate a data model that receives and classifies an inquiry into at least one class in real-time or nearly real-time. One or more actions may be determined based at least in part on the at least one class and presented in a user interface to a user who first submitted the inquiry.

Abstract: Certain aspects of the present disclosure provide techniques for improving a user experience based on electronic records of transactions. Embodiments include training a classifier using training data comprising a set of historical transaction descriptions and a set of corresponding historical classifications that indicate whether or not each historical transaction description of the set of historical transaction descriptions is associated with a user location. Embodiments further include receiving a transaction record describing a transaction associated with a user. Embodiments further include using the classifier to determine a classification for the transaction. The classification indicates whether or not the transaction description is associated with a location of the user. Embodiments further include providing, based at least in part on the classification, a communication to the user that relates to the location of the user.

Abstract: Disclosed herein are systems and methods for replicating data across deployments in a routing constrained environment. To replicate data, a processor may detect a modification that changes data for a source entity within a source environment hosting a source deployment of an application. The processor may then update a target environment hosting a target deployment of the application to mirror the modification within the source environment. To update the target environment, the processor may generate a mapping artifact that identifies the source entity having changed data and the target entity within the target environment receiving the changed data. The processor may then create a mapping infrastructure including one or more compute instances that replicate the changed data for the source entity in the target entity.

Abstract: A computer-implemented system and method for achieving strong consistency in an eventually consistent distributed system. The system comprises a database and an application server hosting a computer software product that a user subscribes to in order to access the product via a user device over a network. The application server receives a first service request from the user device for obtaining a first set of features associated with the user. The application server may ensure that the user receives a useable and correct set of features.

Abstract: Computer-implemented method, tax return preparation systems and computer program products for presenting an explanation for a tax calculation or operation performed by a tax return preparation system when the system is in “forms” mode. An electronic version of a tax authority form that has been modified by inclusion of interface elements is displayed to a user. In response to the user selecting an interface element while in forms mode, presentation of an explanation regarding a tax calculation or operation for the associated field is invoked. The user interface controller provides data in response to selection of the interface element to an explanation engine, which determines an explanation based at least in part upon the tax calculation graph. The explanation is provided to the user interface controller for presentation to the user.

Abstract: A system including a data repository storing metrics describing operational behavior of software programs executing in an enterprise system. The system also includes an application programming interface (API) gateway configured to receive the metrics. The system also includes an ingestion layer configured to ingest the metrics to form the ingested metrics. The system also includes a tumbling window processor configured to process the ingested metrics and the events into heat maps, sort the heat maps into the time slices, and populate the time slices with the ingested metrics.

Abstract: Systems and methods for authenticating a user are disclosed.

Abstract: User data from users/consumers is transformed into machine learning training data including historical offer attribute model training data, historical offer performance model training data, and user attribute model training data associated with two or more users/consumers, and, in some cases, millions, tens of millions, or hundreds of millions or more, users/consumers. The machine learning training data is then used to train one or more offer/attribute matching models in an offline training environment. A given current user's data and current offer data are then provided as input data to the offer/attribute matching models in an online runtime/execution environment to identify current offers predicted to have a threshold level of user interest. Recommendation data representing these offers is then provided to the user and the current user's actions with respect to the recommended offers is monitored and used as online training data.

Abstract: Automatic keyphrase labeling and machine learning training may include a processor extracting a plurality of keywords from at least one search query that resulted in a selection of a document appearing in a search result. For each of the plurality of keywords, the processor may determine a probability that the keyword describes the document. The processor may generate one or more keyphrases by performing processing including selecting each of the plurality of keywords having a probability greater than a predetermined threshold value for insertion into at least one of the one or more keyphrases and assembling the one or more keyphrases from the selected plurality of keywords. The processor may label the document with the keyphrase.

Abstract: A method and system authenticates a user of a data management system. The method and system store an identification tag in the transaction description for a transaction between the user and the data management system. When the user attempts to access the data management system at a later date, the method and system requests that the user provide access to the transaction description. If the user is able to provide access to a transaction description that includes the identification tag, then the method and system can authenticate the user.

Abstract: A method of presenting transactions to a user via a client device for review. Each of the transactions and transactions details are presented in a transaction cell. The user swipes the corresponding transaction cell, thus assigning the associated transaction to a category. In detecting a swipe across a transaction cell, a temporary (yet visible) pause of the swipe interaction is performed, alongside the presentation of an action prompt within the transaction cell directed to performing actions relevant to the transaction. Constrained by a swipe pause timer (SPT), the user is provided a short time window in which to engage the action prompt in order to pursue the actions disclosed by the action prompt. If engagement of the action prompt is not pursued by the user prior to the elapsing of the SPT, the swipe interaction proceeds without hindering the speed with which the user typically reviews a transaction.

Abstract: Systems and methods that may be used to implement overlapping data caching for object application program interfaces (APIs). A unique identifier may be assigned to each sub-component of an object API request. The unique identifiers may be used to determine if sub-components of one object API request overlaps with a sub-component of a prior different request such that a data response for the prior different request can be retrieved from the cache and used as part of the data response for the current object API request.

Abstract: A method collects word-based data corresponding to a first identifier. A first phrase vector is generated for the first identifier by extracting frequency data from the word-based data. A similarity metric is generated corresponding to the first identifier and a second identifier by comparing the first phrase vector of the first identifier to a second phrase vector of the second identifier. A tuple is generated that includes the first identifier and the second identifier using the similarity metric. A machine learning model is trained with the tuple to generate an embedded vector corresponding to the first identifier.

Abstract: Systems and apparatuses are disclosed for extracting information from document images. An example method includes segmenting a document image into multiple segments and determining formatting information for each segment. Determining formatting information for a segment includes determining one or more features of the segment and comparing the one or more features of the segment to one or more clusters of features associated with different document types. The formatting information for the segment is based on the comparison. The method also includes, for each segment, storing the formatting information in a data structure associated with the segment. The method further includes, for each segment including text to be identified during information extraction, applying OCR to the segment to generate machine-encoded text and storing the machine-encoded text in the associated data structure.