The present disclosure relates to techniques for automatically training a chatbot using utterances not understood by the chatbot itself and thus improve its understanding in a fast, effective and unsupervised way. This disclosure implements techniques to update known intents, to create new ones, and to modify the dialog manager so that new responses can be provided. Known intents can be updated with the chatbot's non-understood messages in three different ways: 1) when the user is asked to reformulate his request due to the chatbot not understanding the user, and the confidence of the new intent is greater than a confidence threshold, and the new detected intent is the same as the previous intent; 2) when the user accepts an intent suggestion from the chatbot; and 3) when the agent accepts a response suggestion from the chatbot. New intents, on the other hand, can be automatically created and automatically associated with real answers provided by human agents. These two strategies used simultaneously allow unsupervised training of a dialog system. The benefits of this approach are twofold: 1) there is no need for human intervention to improve the chatbot (unsupervised training), and 2) the new learning introduced in the models comes from real data, not from utterances produced artificially by a human.

The present invention is a system and method of routing incoming communications to a CSR and providing guidance to the CSR based on the incoming communication using feedback information such as sentiment feedback, survey feedback, and feedback from actions taken by CSRs based on skill level. A CEC system receives an incoming communication, analyzes the communication and creates metadata based off of the analysis. The metadata is used by the RAE routing module to route the communication to an appropriate party. The metadata is also used by the GAE guidance module to determine the guidance to provide to the CSR. The CSR then performs an action based on the guidance. The CEC system continues to monitor the interaction until the interaction is completed. The communication metadata, the communication, the guidance, and the CSRs action are all provided to a RAS rules analysis module wherein the RAS analyzes the data and automatedly updates the rules (RAR and GAR) according to the analysis.

A system and method for providing customer-support services to a user of a website through a web widget. When the user accesses a web page containing the web widget, a call is automatically made from the web widget to a customer-support system. In response to the call, the system obtains status information from the customer-support system indicating whether customer-support agents are available to take telephone phone calls. If the status information indicates customer-support agents are available to take telephone calls, the web widget provides a call-related interface to the user, wherein the call-related interface is configured to facilitate a telephone call with a customer-support agent. If the status information indicates customer-support agents are not available to take telephone calls, the web widget provides a non-call-related interface to the user, wherein the non-call-related interface is configured to facilitate an alternative channel of customer support that does not involve a telephone call.

A call management system of a call center identifies an inbound caller based upon computer analysis of customer identifiers, which may include at least two of customer name, street address, and zip code. Approximate string matching analysis matches n-grams generated from strings within customer identifiers, with n-grams generated from customer identification fields while searching one or more databases. Approximate string matching can incorporate a closeness metric based on Jaccard distance, and a Gaussian mixture model of best matches. In one embodiment, a polymr search engine analyzes customer identifiers of inbound callers to retrieve customer data, such as customer demographic data, matched to the customer identifiers. In another embodiment, the polymr search engine analyzes customer identifiers of inbound callers to identify repeat callers and retrieve previously collected customer data. Retrieved customer data is used in predictive modeling and scoring value of the inbound call, and in routing the scored inbound call.

One example method of operation may include determining a call received from a calling party and intended for a subscriber device has an elevated likelihood of being a scam call, determining a percentage of calls over a current period of time being filtered as scam calls by a carrier server, when the percentage of calls being filtered as scam calls during the current period of time is above a call threshold percentage, retrieving call history information associated with a subscriber profile of the subscriber device, identifying one or more call patterns from the call history information of the subscriber profile corresponding to the received call, and determining whether to permit the received call based on the identified one or more call patterns.

A call handler in a contact center is configured to temporarily derogate the skill level of one or more agents in a group having a common skill. During the time period of derogation, incoming calls will be routed to agents with the highest skill level, such that the normally lower skill level agents will be selected to receive calls. If there are more than one agent meeting this criteria, then the agent that has been waiting the longest to receive a call is selected. This allows the lower skill level agents to gain experience in handling calls and improving their call handling capabilities. After a set threshold, the derogation of the agents is terminated, and their skill level is restored to the pre-derogation level. In this manner the lower skill level agents are given limited opportunities for improvement by handling calls.

A call management system of a call center retrieves from a customer database enterprise customer data associated with an identified customer in a customer call, which may include customer event data, attributions data, and activity event data. The customer database tracks prospects, leads, new business, and purchasers of an enterprise. The system retrieves customer demographic data associated with the identified customer. A predictive model is selected from a plurality of predictive models based on retrieved enterprise customer data. The selected predictive model, including a logistic regression model, and tree-based model, determines a value prediction signal for the identified customer, then classifies the identified customer into a first value group or a second value group. The system routes a customer call classified in the first value group to a first call queue assignment, and routes a customer call classified in the second value group to a second call queue assignment.

A method and system for automatically executing an action within a call center environment. Data is aggregated from multiple data sources into a combined data stream. One of the data sources is a source of data corresponding to at least one communication processed by the call center one of the data sources can be a source of external data representing activity that is external to the call center. The combined data stream is processed into successive batches of data corresponding to one or more communications between a call center agent and a communicating party received by the call center. A sensor data structure specifying at least one rule is applied to the batches of data. The at least one rule can include a machine learning model and a configuration data structure based on historical data from the multiple data sources. When it is determined that at least one of the batches satisfies the at least one rule, a notification message relating to the one or more communications is generated. The call center executes an action specified by the sensor data structure based on the notification message. The action can address a situation corresponding the at least one communication. The actions can be various actions such as notifying a specified party, generating an API call, or the like.

Systems and methods for a work distribution service. At a multi-tenant platform that provides a work distribution service for a plurality of external systems, a priority is assigned to a first work item of a first external system. The work item is received via a RESTful work item API call request. The priority is assigned based on work item attributes of the work item and a workflow instruction corresponding to workflow information specified by the work item. The workflow instruction is provided by the external system via a RESTful Workflow API. A worker is assigned to the work item based on: the priority of the work item, the workflow information, and worker state managed by the first external system via a RESTful Worker API. The worker state includes worker attributes. The work item is generated by the external system, and the workflow instruction is managed by the external system.

The technology disclosed herein enables a call to be immediately reconnected to the same agent of a contact center after the agent has been disconnected while the caller remains connected to the contact center. A method provides determining that a first agent system, operated by a first agent of the contact center, has been disconnected from a communication session between the first agent system and a first caller system, operated by a first caller, due to a non-recoverable error. The first caller system remains connected to the contact center. The method further provides generating identification information that identifies the communication session and generating a link to join the communication session that includes the identification information. The method also provides transferring the link to a second agent system operated by the first agent. The second agent system is connected to the communication session when the second agent system follows the link.