Abstract: Systems and methods for using tokenization are provided. In an embodiment, a first computer system receives a notification from a second computer system that user information has changed for a first user account serviced by the second computer system. The notification includes a token corresponding to a second user account serviced by the first computer system. The first user account and the second user account belong to a user. The first computer system searches for the second user account, based on the token, in a database associated with the first computer system. The first computer system updates the second user account to include the user information change and publishes the user information change to other computer systems that subscribe to updates from the first computer system. The published user information change causes the other computer systems to update corresponding user accounts to include the published user information change.

Abstract: There are provided systems and methods for detecting malicious email addresses using email metadata indicators. Digital accounts may be attacked by malicious computing processes or other actions that attempt to compromise the security of accounts and/or perform account takeovers. To increase security of the accounts and account data, the service provider may interface with a digital address and/or identifier provider, such as an email provider to request metadata indications of addresses. The metadata indicator may include a score associated with whether the address is compromised or being used for fraudulent purposes. This score may be based on usages of the address over a period of time, connections of the address, and other activities. The indicator may be used to determine whether to allow data changes to the account's data.

Abstract: There are provided systems and methods for utilizing a vehicle to determine an identity of a user. A user may establish a vehicle as associated with the user, such as a vehicle owned by the user or used by the user. The vehicle may include a vehicle device that may be configured to establish connections with other devices using short range and network communications. In various embodiments, the other devices may be devices at merchant locations. Thus, a merchant at the merchant location may be informed about settings, preferences, and/or parameters for use of the vehicle established by the user. The merchant may receive these parameters even if another user is utilizing the vehicle, thus, establishing rules for use of the vehicle by the other user. In other embodiments, the other devices may be devices in other vehicles, allowing the users to communicate and set waypoints during trips.

Abstract: A system, a medium, and a method are provided to send compensation data to devices at one or more locations where the data is retrievable. For example, a communication interface of a server device accesses request data, where the request data includes an indication of one or more locations where a compensation amount is retrievable. A compensation data component generates compensation data based on the indication of the one or more locations where the compensation amount is retrievable. A location component determines a location of a client device that corresponds to the one or more locations. A transceiver transmits the compensation data to the client device based on the determined location of the client device that corresponds to the one or more locations.

Abstract: A system performs operations that include receiving, via first computing environment, a request to process text data using a first natural language processing (NLP) model. The operations further include accessing configuration data associated with the NLP model, where the configuration data generated using a domain specific language that supports a plurality of preprocessing modules in a plurality of programming languages. The operations also include selecting, based on the configuration data, one or more preprocessing modules of the plurality of preprocessing modules, generating, based on the configuration data, a preprocessing pipeline using the one or more preprocessing modules, and generating preprocessed text data by inputting the text data into the preprocessing pipeline. The preprocessed text data is provided to the first NLP model.

Abstract: There are provided systems and methods for dynamic data query routing for managing access to datasets having multiple versions. A service provider, such as an electronic transaction processor for digital transactions, may provide data for different applications including local device-side software applications and web-based or server-side applications. The data may be read and provided within the application, for example, for output or processing, as well as operated on in the applications to update, add to, or delete from corresponding records. When retrieving or loading data, a data quantum may be used in a data mesh to abstract database physical locations and utilize logical names that allow for applications to code and query for data without direct entanglement with databases and corresponding stored datasets for the data. A router may then be used to route requests and queries for the data to optimized databases for the data.

Abstract: Techniques are disclosed relating to speculatively processing transactions. A transaction processing system may receive an indication of a trigger event associated with an electronic transaction not yet initiated by a client computing device. In some embodiments, in response to the indication of the trigger event and prior to receiving an indication of the electronic transaction being initiated, the computer system begins speculative processing of the electronic transaction. In some embodiments, the speculative processing includes identifying the client computing device based on device authentication information received from the client device, determining a user account based on the identifier client computing device, retrieving account information for the determined account, and storing the retrieved account information.

Abstract: Methods and systems are presented for extracting categorizable information from an image using a graph that models data within the image. Upon receiving an image, a data extraction system identifies characters in the image. The data extraction system then generates bounding boxes that enclose adjacent characters that are related to each other in the image. The data extraction system also creates connections between the bounding boxes based on locations of the bounding boxes. A graph is generated based on the bounding boxes and the connections such that the graph can accurately represent the data in the image. The graph is provided to a graph neural network that is configured to analyze the graph and produce an output. The data extraction system may categorize the data in the image based on the output.

Abstract: A method of categorizing text entries on a document can include determining, for each of a plurality of text bounding boxes in the document, respective text, respective coordinates, and respective input embeddings. The method may further include defining a graph of the plurality of bounding boxes, the graph comprising a plurality of connections among the plurality of bounding boxes, each connection comprising a first and second bounding box and zero or more respective intermediate bounding boxes. The method may further include determining a respective attention value for each connection according to a quantity of intermediate bounding boxes in the connection and, based on a the respective attention values and a transformer-based machine learning model applied to the respective input embeddings and respective coordinates, determining output embeddings for each bounding box and, based on the respective output embeddings, generating a bounding box label for each bounding box.

Abstract: A bank account can be authenticated or confirmed by a user providing an account number and a debit card number. The payment provider then obtains user account information from the bank through a debit card network. The payment provider authenticates the account if the account numbers match. In another embodiment, authentication further requires a portion of a current bank balance in the account provided by the user to match the current balance received from the bank.

Abstract: There are provided systems and methods for configurable storage resource usage through a multi-phase data processing pipeline for data storage efficiency. A service provider may provide online digital computing services, such as electronic transaction processing, account, and the like through online platforms and digital content that may be loaded and served on webpages and through applications of the service provider. The service provider may process and serve data, which may be stored in various data storages and databases of the service provider. The service provider may configure and optimize storage resource usage by processing data through a multi-phased processing pipeline that analyzes the data for reductions in data storage size, duplications, and/or storage times to convert data into a more efficient structure. This may include converting the data to smaller data pieces, deduplicating the same or similar data, or limiting storage in resource intensive data stores.

Abstract: Placing an event into a particular cluster can allow various inferences about the event. A new payment transaction that looks similar to a previously identified cluster of mostly fraudulent payment transactions, for example, may be higher risk. The present disclosure includes structural data improvements to the way that online clustering of events (which may include web events and not just payment transactions) occurs. A new event can be classified into a particular segment very quickly using feature table searching, which can allow for better decision making when a short timeframe is required (e.g. transaction processing, online advertising, etc.).

Abstract: Techniques are disclosed herein relating to using reinforcement learning to generate a dialogue policy. A computer system may perform an iterative training operation to train a deep Q-learning network (DQN) based on conversation logs from prior conversations. In various embodiments, the DQN may include an input layer to receive an input value indicative of a current state of a given conversation, one or more hidden layers, and an output layer that includes a set of nodes corresponding to available responses. During the iterative training operation, the disclosed techniques may analyze utterances from a conversation log and, based on the utterances, use the DQN to determine appropriate responses. Reward values may be determined based on the selected responses and, based on the reward values, the DQN may be updated. Once generated, the dialogue policy may be used by a chatbot system to guide conversations with users.

Abstract: There are provided systems and methods for a compute platform for machine leaning model roll-out. A service provider, such as an electronic transaction processor for digital transactions, may provide intelligent decision-making through decision services that execute machine learning models. When deploying or updating machine learning models in these engines and decision services, a model package may include multiple models, each of which may have an execution graph required for model execution. When models are tested from proper execution, the models may have non-performant compute items, such as model variables, that lead to improper execution and/or decision-making. A model deployer may determine and flag these compute items as non-performant and may cause these compute items to be skipped or excluded from execution. Further, the model deployer may utilize a pre-production computing environment to generate the execution graphs for the models prior to deployment or upgrading.

Abstract: Systems, computer-implemented methods, apparatus, and/or computer program products that can facilitate electronic generation of enhanced, transaction-based QR codes for improved fraud detection are provided. In various aspects, a processor can receive a signal from an electronic beacon of a point-of-sale device, wherein the signal includes information associated with a first merchant. In various instances, in response to receiving the signal from the electronic beacon, the processor can identify a quick response (QR) code from a set of quick response codes that corresponds to the first merchant. In various cases, the processor can render the identified QR code on an electronic display that is scannable by the point-of-sale device, wherein scanning the identified QR code causes the point-of-sale device to transmit the identified QR code and a second merchant identifier to a payment processor.

Abstract: Methods and systems are presented for providing automated online chat assistance in an online chat session. One or more utterances transmitted from a user device of a user via the online chat session are obtained. The one or more utterances are provided to a first prediction model to predict an intent of a user. If it is determined that the first prediction model is unable to predict the intent of the user based on the one or more utterances, the one or more utterances are provided to a second prediction model. After predicting the intent of the user by the second prediction model, the intent is used by a chat robot to provide a dialogue with the user via the online chat session. The one or more utterances and the predicted intent are used to re-train the first prediction model.

Abstract: Disclosed methods include maintaining a database of resource limits for a plurality of agents. A resource limit may be usable for predicting a result of a given request from a given agent. Maintaining the database may include determining an updated resource limit for a particular agent based on identifying an extrema point of a function of resource limit. The maintaining may further include updating the database using the updated resource limit, as well as selecting, from the database, a subset of the plurality of agents that are selected based on associated parameter values compared to parameter values associated with the particular agent. The maintaining may also include updating corresponding resource limits for the subset of the plurality of agents based on the updated resource limit. The method may further include receiving a request from the particular agent, and predicting, using the updated resource limit, a result of the request.

Abstract: Methods and systems are presented for auditing user feedback data corresponding to user communications received via at least one interface of a service provider. The user feedback data includes a first set of feedback categories associated with a first classification of the user communications. A first feature representation of the communications is generated from the user feedback data. The first feature representation includes a first set of textual data features extracted from the communications. A second feature representation is generated from the first feature representation using a first machine learning model. The second feature representation includes a second set of textual data features including semantic equivalents of the first set of features. A second machine learning model is trained with the second feature representation. A second classification of the user communications according to a second set of feedback categories is generated using the trained second machine learning model.

Abstract: Character string analysis and classification can be useful in a variety of contexts, including examining web URLs to determine whether a URL indicates that a user is attempting to take a particular action on an electronic service platform. In some cases, however, URLs or other string data may have “noise” in them, such as random sub-strings, that prevents a string from being properly classified. Sometimes it may be useful to classify a string into a category, however, and it may be important to do this quickly (e.g. during an active user interaction with a website). Learning tables allowing for O(1) lookup can be established by tokenizing strings and then using probability analysis to eliminate tokens that appear an insufficient number of times. This allows for quick and accurate string classification, which may be useful in numerous circumstances.

Abstract: Methods and systems are presented for providing a container orchestration framework for facilitating development and deployment of software applications across different operating environments within an enterprise system. Upon receiving a service request for processing a set of data is received, the container orchestration framework determines one or more machines that store the set of data. Instead of processing the set of data remotely, the container orchestration framework deploys a container that encapsulates an application on the one or more machines. Each application instance running on the one or more machines are executed to process a corresponding subset of data stored on the machine locally. The container orchestration framework obtains the output data from executing the applications on each of the one or more machines, and present the output data as a response to the service request.