Abstract: Methods, systems and computer program products implementing hierarchical classification techniques are disclosed. A hierarchical classification system receives training data including labeled transaction records. The system determines tag sequences from the training data. The system clusters the tag sequences into clusters. The system determines a cluster-level classifier that is trained to predict a cluster for an input transaction record. The system determines a respective cluster-specific classifier for each cluster. The system trains the cluster-specific classifier to predict a label of entity of interest for an input transaction record, given a particular cluster. Upon receiving a test transaction record, the system first applies the cluster-level classifier to determine a particular cluster for the test transaction record, and then determines a label of entity of interest of the test transaction record by applying a cluster-specific classifier of that particular cluster.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for receiving, from a user device, a request to include financial data describing a financial account in an interface, the financial account being associated with a particular financial institution; identifying one or more metadata attributes for accessing the financial account associated with the particular financial institution; obtaining a public digital key associated with the financial institution; providing, to the user device, data describing the one or more metadata attributes and the public digital key associated with the financial institution; receiving, from the user device, respective values for the one or more metadata attributes, the corresponding values having been encrypted by the user device using the public digital key; and storing the encrypted values for the one or more metadata attributes for use in accessing and aggregating financial data describing the financial account.

Abstract: Methods, systems and computer program products implementing character-level deep neural networks for information extraction are disclosed. A system uses character-level information retrieved from a transaction record to classify the transaction as a whole and to tag individual sections of the transaction record by entity type. The system processes the transaction record using multiple and separate character-level models. The system can use a one-dimensional neural network for featurization fed into a fully connected network for classification for identifying the most common classes of a transaction record. The system can identify one or more entities, e.g., service provider names, from the transaction using an RNN. The RNN can include one or more LSTM models. The LSTM models can be BI-LSTM models.

Abstract: Methods, systems and computer program products generating diverse and representative set of samples from a large amount of transaction data are disclosed. A data sampling system receives transaction records. Each transaction record has multiple text segments. The system selects a subset of transaction records that contain least frequently appeared text segments. The system determines a respective vector representation for each selected transaction record. The system can measure similarity between transaction records based on the vector representations. The system assigns the selected transaction records to multiple clusters based on the vector representations and designated dimensions of importance. The system identifies one or more anchors that include transaction records on boundaries between clusters. The system filters the subset of transaction records by removing transaction records that are close to the anchors.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for account authentication. A method includes receiving a user request to include financial data describing a financial account in an interface, the financial account being associated with a financial institution. The method further includes redirecting the user to a first webpage associated with the financial institution, where the user inputs into the first webpage login credentials for accessing the financial account. The method further includes, in response to the user inputting into the first webpage login credentials for accessing the financial account, receiving, from the financial institution, an access token other than the login credentials for accessing the financial account. The method further includes storing the access token for use in accessing and aggregating financial data describing the financial account.

Abstract: Methods, systems and computer program products implementing hierarchical classification techniques are disclosed. A hierarchical classification system receives training data including labeled transaction records. The system determines tag sequences from the training data. The system clusters the tag sequences into clusters. The system determines a cluster-level classifier that is trained to predict a cluster for an input transaction record. The system determines a respective cluster-specific classifier for each cluster. The system trains the cluster-specific classifier to predict a label of entity of interest for an input transaction record, given a particular cluster. Upon receiving a test transaction record, the system first applies the cluster-level classifier to determine a particular cluster for the test transaction record, and then determines a label of entity of interest of the test transaction record by applying a cluster-specific classifier of that particular cluster.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for data aggregation.

Abstract: Methods, systems and computer program products for layered masking of data are described. A system receives content including personally identifiable information (PII). The system redacts the content by masking the PII. The system identifies the PII in multi-layer processing, where in each layer, the system determines a respective confidence score indicating a probability that a token is PII. If the confidence score is sufficiently high, the system masks the token. Otherwise, the system provides the token to a next layer for processing. The layers can include regular expression based processing, lookup table based processing, and machine learning based processing.

Abstract: Methods, systems and computer program products implementing data aggregation using distributed Web scraping are disclosed. A mobile device can scrape one or more target sites to collect data from accounts of a particular user. The scraping can occur under scraping conditions as specified by the user. The scraping conditions can include conditions based on time, power, bandwidth, usage, or any combination of the above. The scraping conditions can ensure that the scraping occurs at time that is most convenient to the user, e.g., when sufficient bandwidth is available to the mobile device or the mobile device is not performing other tasks. The mobile device can upload the scraped data to a data aggregation server under submission conditions as specified by the user. The data aggregation server can aggregate the scraped data, enrich the aggregated data, and provide the enriched data to the user through Web access.

Abstract: Methods, systems and computer program products for identifying recurring series from transactional data are disclosed. An analytic server receives transactional data. The server reduces the transactional data to one or more series of transactions. The analytic server identifies one or more features in each series of transactions. The analytic server determines a respective recurrence score for each series of transactions based on training data. The recurrence score measures a likelihood that the series of transactions is a recurring series, in which transactions recur. The analytic server determines whether each series of transactions is a recurring series based on the respective recurrence score and a threshold. The analytic server predicts a future transaction based on the recurring series of transactions.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for decoding Turing tests. One of the methods includes managing a database that stores data of each of a plurality of aggregation accounts; sending, for a particular account identified by one of the aggregation accounts and to a server, a request for access to account data for the particular account; receiving, from the server, data that includes a login credentials field and a Turing test challenge; extracting the Turing test challenge; providing, to an external system that is a different system from the server, the Turing test challenge; receiving, from the external system, a response to the Turing test challenge; providing, to the server, the response to the Turing test challenge; providing, to the server, the login credentials for the particular account; and receiving, from the server, account data for the particular account.

Abstract: Methods, computer program products, and systems, related to providing financial gadgets to users. In one implementation, a method includes receiving gadgets and receiving a notification that one or more gadgets has been certified. A web page with an interface through which users can select gadgets is generated and provided to a user computer. A user selects a gadget, and the selected gadget is run, including providing the user-specific financial information to the gadget. Content based on the output of the gadget is provided to the user. User-specific financial information can be collected, and modified by the gadget. A notification can be received that a financial institution has authorized one or more of the certified gadgets and the web page can only allow users to select authorized gadgets.

Abstract: Methods, systems and computer program products for layered masking of data are described. A system receives content including personally identifiable information (PII). The system redacts the content by masking the PII. The system identifies the PII in multi-layer processing, where in each layer, the system determines a respective confidence score indicating a probability that a token is PII. If the confidence score is sufficiently high, the system masks the token. Otherwise, the system provides the token to a next layer for processing. The layers can include regular expression based processing, lookup table based processing, and machine learning based processing.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for receiving, from a user device, a request to include financial data describing a financial account in an interface, the financial account being associated with a particular financial institution; identifying one or more metadata attributes for accessing the financial account associated with the particular financial institution; obtaining a public digital key associated with the financial institution; providing, to the user device, data describing the one or more metadata attributes and the public digital key associated with the financial institution; receiving, from the user device, respective values for the one or more metadata attributes, the corresponding values having been encrypted by the user device using the public digital key; and storing the encrypted values for the one or more metadata attributes for use in accessing and aggregating financial data describing the financial account.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for account authentication. A method includes receiving a user request to include financial data describing a financial account in an interface, the financial account being associated with a financial institution. The method further includes redirecting the user to a first webpage associated with the financial institution, where the user inputs into the first webpage login credentials for accessing the financial account. The method further includes, in response to the user inputting into the first webpage login credentials for accessing the financial account, receiving, from the financial institution, an access token other than the login credentials for accessing the financial account. The method further includes storing the access token for use in accessing and aggregating financial data describing the financial account.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for data aggregation.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for decoding Turing tests. One of the methods includes managing a database that stores data of each of a plurality of aggregation accounts; sending, for a particular account identified by one of the aggregation accounts and to a server, a request for access to account data for the particular account; receiving, from the server, data that includes a login credentials field and a Turing test challenge; extracting the Turing test challenge; providing, to an external system that is a different system from the server, the Turing test challenge; receiving, from the external system, a response to the Turing test challenge; providing, to the server, the response to the Turing test challenge; providing, to the server, the login credentials for the particular account; and receiving, from the server, account data for the particular account.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for automating the collection of user information for account aggregation. In one aspect, a method includes receiving, at a server computer system from a mobile device of a user, a first user request to access account information; receiving, at the server computer system from a provider computer system, a plury of mobile device applications; for each respective establishment of the plurality of establishments: storing, at the server computer system, establishment login credentials of the user to access account information of the user at a computer system of the respective establishment, and obtaining account information of the user at the respective establishment; aggregating, on the computer system, all the account information of the user from the respective mobile device application of each establishment; and providing to the mobile device the aggregated account information of the user.

Abstract: A computer-implemented method includes issuing a transaction on the Automated Clearing House (ACH) electronic network, the transaction including instructions for printing a check, receiving the transaction, and printing the check according to the instructions.

Abstract: Methods, systems and computer program products for identifying recurring series from transactional data are disclosed. An analytic server receives transactional data. The server reduces the transactional data to one or more series of transactions. The analytic server identifies one or more features in each series of transactions. The analytic server determines a respective recurrence score for each series of transactions based on training data. The recurrence score measures a likelihood that the series of transactions is a recurring series, in which transactions recur. The analytic server determines whether each series of transactions is a recurring series based on the respective recurrence score and a threshold. The analytic server predicts a future transaction based on the recurring series of transactions.