Invented by Yochai Konig, David Konig, Bank of America NA
The Bank of America NA invention works as follows
The method includes selecting a topic to interact with a human contact center agent, identifying a dialog-tree associated with that topic, and engaging in an automated communication with the human contact center agent using the identified dialog-tree.Background for Chat bots for automatic quality management of chat agents
Modern contact centers, in general, are staffed by agents or employees that serve as the interface between an organisation, like a company, with external entities, like customers. Human sales agents in contact centers can help customers make purchasing decisions, and they may also receive orders from customers. Human support agents in contact centers can help customers with problems relating to products and services offered by an organization. Contact center agents may interact with outside entities, such as customers, through speech-voice (e.g. telephone calls, voice over IP, or VoIP), video (e.g. video conferencing), or text (e.g. emails, text chats, or text messages).
Quality monitoring is the process by which contact centers evaluate agents to ensure that they are providing high-quality service when assisting customers. A quality monitoring process monitors the performance of agents by evaluating their interactions with customers. This includes assessing whether they were polite, courteous, efficient and knowledgeable.
The information in the Background section may not be prior art, but is meant to enhance understanding of the background.
One or more aspects” of an example embodiment of the present invention relate to a system and a method of utilizing an automated chat program in order to simulate human customers for quality management purposes, while interacting with human agents.
The following is an example of an embodiment of a method to automate quality management for agents in a contact centre: selecting a topic to interact with a human contact center agent, identifying a dialog-tree associated with that topic, and engaging the human contact center agent in an automated communication based on this identified dialog-tree. Receiving an input from the agent, the processor identifies a node in the dialog tree associated to the input, selecting an automated phrase in response to the node, and then outputting the phrase.
The method can also include: determining whether the agent’s input is semantically equal to that of the target input by the processor; calculating a score for the human agent on the basis of the determining; and displaying feedback to the user based upon the calculated score.
In one embodiment, feedback could include a summary of the agent’s strengths and/or weaknesses.
In one embodiment, the method can also include the following: the processor may invoke a coaching session with the human agent on the basis of the feedback.
In one embodiment, the automated phrases may be selected from the plurality of phrases provided to agents by customers during a current dialog status, relating to a selected topic, in interactions with the contact center.
In one embodiment, selecting the automated phrase can further include: identifying by the processor the frequency of each of a plurality phrases for the dialog state in use; and selecting by the processor one of the plurality based on that identified frequency.
In one embodiment, the contact center may select the topic based on an optimization criterion.
In one embodiment, the selected topic is based on the performance of the agent in interactions with customers who are interested in the topic.
In one embodiment, the topic can be chosen based on the performance of the human agent in previous automated communication sessions.
In one embodiment, an automated communication session can be a chat session using text.
The system is based on a dialog tree that has been identified as being associated with a selected topic. It also includes a processor and a memory. When executed by the processor the instructions cause the processor: to select a topic to interact with a human contact center agent; to identify the dialog tree; to engage in an automated conversation with the agent using the dialog tree; to select an automated phrase in response to identifying this node, and to output the phrase.
The instructions can further instruct the processor to perform the following: identify a target input for the human agent associated with the current node identified; compare the input of the agent to that of the target input; determine if the input is semantically equal to the input of the target agent; calculate a score based on this determining; then output feedback based upon the calculated score.
In one embodiment, feedback could include a summary of the agent’s strengths and/or weaknesses.
In one embodiment, the instructions could also cause the processor: to invoke a coaching for the human agent on the basis of the feedback.
In one embodiment, the automated phrases may be selected from the plurality of phrases provided to agents by customers during a current dialog status, relating to a selected topic, in interactions with the contact center.
In one embodiment, selecting the automated phrase can further include identifying the frequency of each of a plurality phrases for the dialog state currently in use; and selecting a phrase from the plurality based on that identified frequency.
In one embodiment, the contact center may select the topic based on an optimization criterion.