Abstract: A system for data recording across a network includes a session border controller connecting incoming data from the network to an endpoint recorder. A load balancer is connected to the network between the session border controller and the endpoint and receives the incoming data from the session border controller, wherein the load balancer comprises computer memory and a processor configured to parse the incoming data into video data and audio data according to identification protocols accessible by the processor from the computer memory. A recording apparatus includes recording memory that receives the incoming data from the load balancer, stores a duplicate version of the incoming data in the recording memory, and connects the incoming data to the endpoint.

Abstract: To allow the human customer service agents to specialize in the instances where human service is preferred, but to scale to the volume of large call centers, systems and methods are provided in which human agents and intelligent virtual assistants (IVAs) co-handle a conversation with a customer. IVAs handle simple or moderate tasks, and human agents are used for those tasks that require or would benefit from human compassion or special handling. Instead of starting the conversation with an IVA and then escalating or passing control of the conversation to a human to complete, the IVAs and human agents work together on a conversation.

Abstract: An anomaly detection system using machine learning to generate predicted survey scores for a given duration and a given metric based on historic survey score data. The system compares a predicted survey score to the actual survey score and identifies anomalous actual survey scores. The anomaly detection system trains a plurality of survey score prediction models using historic survey score data. Each survey score prediction model is based on a specific survey score metric and a specific duration. The survey score prediction models generate expected survey score results for the given duration and the given metric. Based on the user-determined filtering and tolerances, the system determines if the actual survey score result is anomalous. The system generates reports for the detected anomalies and continually updates the survey score prediction models with newly obtained actual survey results, thereby improving the anomaly detection accuracy over time.

Abstract: A system and method for updating computerized language models is provided that automatically adds or deletes terms from the language model to capture trending events or products, while maximizing computer efficiencies by deleting terms that are no longer trending and use of knowledge bases, machine learning model training and evaluation corpora, analysis tools and databases.

Abstract: The template generation system receives interaction data stored by the CEC from an interaction database and customer service templates (if any) from a template database. The template generation system processes interaction data and customer service templates to learn the domain language of CSR responses and the template responses within the CEC. The template generation system encodes the learned language and generates sentence vector embeddings for the CSR responses and template responses. Based on the learned language, the encoding, and the sentence vector embeddings, the template generation system processes CSR responses derived from the interaction data and customer service templates to predict the need for new customer service templates. Based on the predicted need for new customer service templates, the template generation system provides customer service template suggestions and may also auto-generate suggested customer service templates.

Abstract: Certain aspects of the present disclosure provide techniques for receiving, from a user, a command for a virtual assistant to perform a task on behalf of the user; determining a communication channel for the virtual assistant to communicate with a remote service in order to perform the task; registering a communication session with an identity provider service, wherein the communication session is associated with the communication channel; initiating the communication session with the remote service using the communication channel; receiving a communication session authentication query from the remote service; and determining, in response to the communication session authentication query, whether the user is authenticated.

Abstract: Systems and methods are provided for dynamically adjusting a website of an entity using information that has been received, stored, gathered, and/or otherwise obtained about what people want to find on the entity's website. A website may be dynamically adjusted using trending information in response to determining that the usage of the monitored data source is greater than the baseline usage distribution or in response to determining that the usage of the monitored data source is not greater than the baseline usage distribution receiving NLP inputs of the user from the IVA and adjusting dynamic web content displayed on the website based on the NLP inputs.

Abstract: System and method for calibration of WFM system modeling parameters. A first mode M[D,S] of a modeler computes demand-shrinkage controlled service levels and an error metric e(M[D,S]) between the controlled and actual service levels. A user device iteratively adjusts each core parameter. When the user is satisfied that e(M[D,S]) is sufficiently small, calibration of the core parameters is complete. The same is done for calibrating the modeling factor, and then a final e(M[D,S])f is computed. A second mode M[D] computes, using the parameters just calibrated, demand-controlled service levels and an error metric e(M[D]) between the controlled service levels and actual levels. The user iteratively adjusts the shrinkage. When the user is satisfied that e(M[D]) is sufficiently small, calibration of the core parameters is complete.

Abstract: A system and method use a trained transformer model to generate summaries of audio interactions based on keywords. Training the transformer model includes obtaining a transcription of an audio interaction, obtain keywords for summarizing the audio interaction, training a transformer model to generate a summary of the audio interaction based on the keywords and the transcription, where the transcription is an input to the transformer model and the keywords are injected between an encoder and a decoder of the transformer model, and deploying the trained transformer model to be used for generating summaries of subsequent audio interactions.

Abstract: The present invention allows a CEC system to automatedly track the use, storage, access, and modification of sensitive information/data in the system through desktop monitoring. Further, through desktop, video, and audio monitoring of CSRs the system can automatedly determine the improper use, access, storage, and modification of sensitive information by implementing sensitive data use rules that allow a system to be specialized for the user. Finally, the system can automatedly determine and implement violation actions for the improper use, storage, access, and manipulation of sensitive information. This provides an intelligent system capable of locating all sensitive data in the system and regulating the use, access, and storage of sensitive data in the system.

Abstract: Systems and methods are described to combine two or more linear models into a combined linear model. Two or more linear models and an observation of interest are selected. The linear models are concurrent with respect to the observation of interest. The observation of interest includes a class value and a feature vector. A distance is selected, and a plurality of feature vectors are selected that are within the distance of the feature vector associated with the observation of interest. These feature vectors are input to the selected linear models and a plurality of class values are generated. These class values and the selected feature vectors are used to generate a combined linear model. The combined model is either a mean local linear surrogate model or a linearly weighted local surrogate model.

Abstract: An interactive voice response (IVR) system including iterative speech recognition with semantic interpretation is deployed to process an audio input in a manner that optimizes and conserves computing resources and facilitates low-latency discovery of start-of-speech events that can be used to support external processes such as barge-in operations. The IVR system can repeatedly receive an audio input at a speech processing component and apply an input-aware recognition process to the audio input. In response to generating a start-of-speech event, the IVR system can apply an input-unaware recognition process to the remaining audio input and determine a semantic meaning in relation to the relevant portion of the audio input.

Abstract: Conversation user interfaces that are configured for virtual assistant interaction may include contextual interface items that are based on contextual information. The contextual information may relate to a current or previous conversation between a user and a virtual assistant and/or may relate to other types of information, such as a location of a user, an orientation of a device, missing information, and so on. The conversation user interfaces may additionally, or alternatively, control an input mode based on contextual information, such as an inferred input mode of a user or a location of a user. Further, the conversation user interfaces may tag conversation items by saving the conversation items to a tray and/or associating the conversation items with indicators.

Abstract: An architecture for assessing and identifying fraudulent contact with client contact systems, such as IVR, includes threshold and machine learning scoring and filtering of calls based on these criteria. The criteria may include behavioral, situational and reputational scoring.

Abstract: A tagging system gathers all events (tagged and untagged) generated by remote sensors at a location or facility over time. Based on the gathered events the tagging system uses machine learning to train a model to learn the sensor layout of a facility or location and the timing between the triggering of sensors. Once trained, the model can predict the movement and location of individuals and objects throughout the facility based on a starting tagged event. Given a series of tagged and untagged events, the system can use the movement predictions of the model to tag the untagged events in the series with the identification of an individual or object that triggered the generation of the untagged event.

Abstract: The present invention allows a CEC system to automatedly track the use, storage, access, and modification of sensitive information/data in the system through desktop monitoring. Further, through desktop, video, and audio monitoring of CSRs the system can automatedly determine the improper use, access, storage, and modification of sensitive information by implementing sensitive data use rules that allow a system to be specialized for the user. Finally, the system can automatedly determine and implement violation actions for the improper use, storage, access, and manipulation of sensitive information. This provides an intelligent system capable of locating all sensitive data in the system and regulating the use, access, and storage of sensitive data in the system.

Abstract: Disclosed are a framework and method for selecting an anomaly detection method for each of a plurality of class of time series based on characteristics a time series example that represents an expected form of data. The method provides classification of a given time series into one of known classes based on expected properties of the time series, filtering the set of possible detection methods based on the time series class, evaluating the remaining detection methods on the given time series using the specific evaluation metric and selecting and returning a recommended anomaly detection method based on the specific evaluation metric.

Abstract: To allow the human customer service agents to specialize in the instances where human service is preferred, but to scale to the volume of large call centers, systems and methods are provided in which human agents and intelligent virtual assistants (IVAs) co-handle a conversation with a customer. IVAs handle simple or moderate tasks, and human agents are used for those tasks that require or would benefit from human compassion or special handling. Instead of starting the conversation with an IVA and then escalating or passing control of the conversation to a human to complete, the IVAs and human agents work together on a conversation.

Abstract: A non-ontological hierarchy for language models is based on established psycholinguistic and neuro-linguistic evidences. By using non-ontological hierarchies, a more natural understanding of user's inputs and intents improve toward a better potential for producing intelligent responses in a conversational situation.

Abstract: Systems and methods are described to automatically generate candidate questions and responses to speed the process of response creation and editing for commercial IVAs and chatbots. Rather than create the questions and responses from scratch for a new IVA, the system uses existing questions and responses from a previous or related IVA to train a model that can generate proposed responses to provided questions. The model, or a different model, can further be trained to generate responses using data taken from company or entity-specific data sources such as websites and knowledge bases. After a set of questions and responses have been generated for an IVA they may be reviewed by one or more human reviewers to ensure they are of a suitable quality. Where no previous or related IVA exists to provide example responses, the model may be trained solely using the company or entity-specific data.