Abstract: Disclosed in some examples are methods, systems, and machine readable mediums which provide summaries of topics determined within a corpus of documents. These summaries may be used by customer service associates, analysts, or other users to quickly determine both topics discussed and contexts of those topics over a large corpus of text. For example, a corpus of documents may be related to customer complaints and the topics may be summarized to produce summaries such as “credit report update due to stolen identity.” These summarizations may be used to efficiently spot trends and issues.

Abstract: An automated teller machine (“ATM”) receives, from a user computing device via a first wireless communication session, a unique transaction identifier (“UTN”). The ATM transmits the UTN to a bank computing system via a network. The ATM receives, from the bank computing system responsive to transmitting the UTN, transaction details and a first security code associated with the UTN. The ATM receives, from the user computing device via a second wireless communication session, a second security code. Responsive to determining that the first security code matches the second security code, the ATM processes a transaction in accordance with the transaction details.

Abstract: A voice communications computer system (“VCCS”) receives a ring signal from a call device having unverified device identification data. The VCCS identifies an audible frequency component and an electronic frequency component of the ring signal. The VCCS identifies a device identification characteristic or a geographic location characteristic based on the audible or electronic frequency components, and identifies a stored identification characteristic associated with the device identification data. Based on a comparison of the stored identification characteristic with the device identification characteristic or geographic location characteristic, the VCCS generates fraud estimation data. In some cases, the VCCS generates call status data based on the fraud estimation data. The VCCS provides the fraud estimation data or the call status data to a user interface device, which is configured to display data or perform a call action for a call associated with the ring signal.

Abstract: Techniques are described for determining a likelihood that a customer communication is fraudulent using one or more machine learning models. For example, a computing system includes a memory and one or more processors in communication with the memory. The one or more processors are configured to: receive a set of emotion factor values for communication data of a current communication associated with a customer, wherein each emotion factor value indicates a measure of a particular emotion factor in the current communication; classify, using an emotion variance model running on the one or more processors, the current communication into an emotional fraud category based on the set of emotion factor values for the current communication associated with the customer; and determine a risk score for the current communication indicative of a probability that the current communication is fraudulent based on at least the emotional fraud category for the current communication.

Abstract: Systems, apparatuses, methods, and computer program products are disclosed for predicting an emotion based on speech and text derived from the speech. An example method includes capturing speech and generating hidden vectors based on the speech and text derived from the speech. The example method further includes generating a word aligned hidden vector based on the speech hidden vector and text hidden vector. The example method further includes generating a hidden vector based on the word aligned hidden vector and the text hidden vector. The example method further includes, after further processing of the final vector, generating one or more probabilities corresponding to one or more emotions. The example method finally includes determining a next action based on the generated one or more probabilities corresponding to one or more emotions.

Abstract: Techniques are described for determining a likelihood that a customer communication is fraudulent using one or more machine learning models. For example, a computing system includes a memory and one or more processors in communication with the memory. The one or more processors are configured to: receive a set of emotion factor values for communication data of a current communication associated with a customer, wherein each emotion factor value indicates a measure of a particular emotion factor in the current communication; classify, using an emotion variance model running on the one or more processors, the current communication into an emotional fraud category based on the set of emotion factor values for the current communication associated with the customer; and determine a risk score for the current communication indicative of a probability that the current communication is fraudulent based on at least the emotional fraud category for the current communication.

Abstract: Disclosed in some examples are methods, systems, and machine readable mediums which provide summaries of topics determined within a corpus of documents. These summaries may be used by customer service associates, analysts, or other users to quickly determine both topics discussed and contexts of those topics over a large corpus of text. For example, a corpus of documents may be related to customer complaints and the topics may be summarized to produce summaries such as “credit report update due to stolen identity.” These summarizations may be used to efficiently spot trends and issues.

Abstract: Sharing of user data of customers of a first party with a third party can be monitored. The data can be presented to customers to enable transparency with respect to what data is provided to whom. Furthermore, remediation can be promptly triggered in response to a third-party data breach. After breach detection, customers and data affected by the breach can be determined. The type of remediation can be determined based on the risk as determined based on the customers affected and the data involved.

Abstract: A voice communications computer system (“VCCS”) receives a ring signal from a call device having unverified device identification data. The VCCS identifies an audible frequency component and an electronic frequency component of the ring signal. The VCCS identifies a device identification characteristic or a geographic location characteristic based on the audible or electronic frequency components, and identifies a stored identification characteristic associated with the device identification data. Based on a comparison of the stored identification characteristic with the device identification characteristic or geographic location characteristic, the VCCS generates fraud estimation data. In some cases, the VCCS generates call status data based on the fraud estimation data. The VCCS provides the fraud estimation data or the call status data to a user interface device, which is configured to display data or perform a call action for a call associated with the ring signal.

Abstract: Sharing of user data of customers of a first party with a third party is monitored. The data is presented to customers to enable transparency with respect to what data is provided to whom. Furthermore, remediation is promptly triggered in response to a third-party data breach. After breach detection, customers and data affected by the breach is determined. The type of remediation is determined based on the risk as determined based on the customers affected by the data involved.

Abstract: Systems, apparatuses, methods, and computer program products are disclosed for predicting an entity, intent, and emotion based on weighted and segmented portions of captured speech. An example method includes capturing speech and separating the speech and/or text derived from the speech into a plurality of discrete segments. The example method further includes determining or generating metadata, the metadata an entity, an intent, an emotion, and a weight associated with each of the entity, the intent, the emotion. The example method further includes determining a next action based on the determined or generated entity, intent, and emotion of each of the plurality of discrete segments.

Abstract: A voice communications computer system (“VCCS”) receives a ring signal from a call device having unverified device identification data. The VCCS identifies an audible frequency component and an electronic frequency component of the ring signal. The VCCS identifies a device identification characteristic or a geographic location characteristic based on the audible or electronic frequency components, and identifies a stored identification characteristic associated with the device identification data. Based on a comparison of the stored identification characteristic with the device identification characteristic or geographic location characteristic, the VCCS generates fraud estimation data. In some cases, the VCCS generates call status data based on the fraud estimation data. The VCCS provides the fraud estimation data or the call status data to a user interface device, which is configured to display data or perform a call action for a call associated with the ring signal.

Abstract: Quality associated with an interpretation of data captured as unstructured data can be determined. Attributes can be identified within the unstructured data automatically. Subsequently, sentiment associated with each of the attributes can be determined based on the unstructured data. Correctness of the unstructured data, and thus the interpretation, can be assessed based on a comparison of the attribute and associated sentiment with structured data. A quality score can be generated that captures the quality of the data interpretation in terms of correctness and as well as results of another analysis including completeness, among others. Comparison of the quality score to a threshold can dictate whether or not the interpretation is subject to further review.

Abstract: Out of office management is provided to an institution and customers of the institution. Employee data can be monitored of one or more employees of the institution. A likelihood an employee is out of office is determined. A set of projects of the employee that are affected by the out of office is determined. The projects are analyzed to determine departure scores for each project of the set of projects. The projects are ranked according to the departure scores to determine criticality of each project. The projects are reassigned based on the employee data and the ranking. Customer sentiments are monitored after reassignment for machine learning to affect future out of office reassignments.

Abstract: Automatic decisioning associated with unstructured data is disclosed. Unstructured data, such as that associated with comments of an underwriter regarding a credit decision, can be received. Text mining can be performed to extract features from the unstructured data. The extracted features can subsequently be provided as input to a machine learning model configured to return a prediction of a class associated with the unstructured data. The predicted class, such as approved or rejected, can subsequently be conveyed for display on a display device.

Abstract: Systems and methods that facilitate determining and predicting fraudulent or non-compliant behavior using poly-structured data analytics are discussed. An unstructured data steam comprising a set of emails can be received, and processed to reduce the noise, or non-relevant portions of the dataset. Structured data that includes contextual relevant information about the users can also be received, and the poly-structured data modeling system can identify unstructured variable and structured variables that are relevant for identifying and predicting non-compliant behavior. These unstructured and structured variables can then be modeled in order to identify communications that may contain non-compliant and/or fraudulent behavior.