Systems and methods relating to emotion-based action recommendations
A system and method are presented relating to emotion-based action recommendations. The present invention may include a recommender engine that analyzes several pieces of information and using artificial intelligence derived models, provides a course of action for a contact center agent and the probability and estimations behind the recommendation as a justification. System inputs include cost of the item or service being troubleshooted, estimated costs for different resolutions, cost of each escalation level of the interaction, history of previous interactions, agent costs, history of the current interaction, acoustic events and sentiment or primary detected emotions of the customer's transcript. The recommended course of action, rationale, and authorization can be provided as output. The system may record whether the agent followed the recommendation to improve further recommendations.
1. A method for automatically providing recommendations to an agent in a contact center environment, the method comprising the steps of:
receiving a voice interaction in the contact center from a first party about an issue;
retrieving information from a record associated with the first party, wherein the information comprises at least one of:
previous orders,
total value of the orders,
cost of goods sold,
cost to return an item,
interaction history of the first party with the contact center,
end-call codes from each interaction in the history of the first party,
sentiment analysis of historical interaction transcripts, and
manually input information from an agent of the contact center;
generating a transcription of the voice interaction in real-time through a speech recognition service;
analyzing the transcription for named entity recognition, topic identification, and sentiment analysis for each utterance of the first party to automatically determine the issue behind the interaction and determining a primary emotion among a plurality of emotions of the first party from the transcription by classifying the sentiment analysis using a neural network with LSTM units;
matching an action to the emotion of the first party where each action is associated with an intent, and the interaction is represented by a series of data points associated with intent of the interaction;
predicting customer disposition and reception to a course of resolution using the series of data points associated with the intent to resolve the issue through a plurality of means, considering a specified business priority comprising one of reducing average handling time or increasing a rate of first call resolution, wherein those means comprise at least one of:
troubleshooting,
refund,
return, and
reimbursement; and
returning a suggested course of action to the agent that maximizes the business priority.
2. The method of claim 1 , wherein the method further comprises the step of recording data on whether a course of action followed by the agent followed the suggested course of action and inputting the recorded data into the system as a feedback loop for improving future suggested courses of action.
3. The method of claim 1 , wherein the determining comprises using an algorithm formulated based on sequence classification from contact center historical data.
4. The method of claim 3 , wherein the contact center historical data further comprises: product loss, labor cost, cost of escalations, cost of public relations, interaction cost, and potential loss of future sales.
5. The method of claim 1 , wherein inputs to the neural network comprise at least one of: sentiment analysis of the transcription, call tracking flags, mood analytics, speech analytics, image-based motion detection, and end-call codes.
6. The method of claim 1 , wherein the suggested course of action comprises a rationale for the suggestion and an authorization for the agent to conduct said course of action.
7. A method for automatically providing recommendations to an agent in a contact center environment, the method comprising the steps of:
receiving a chat interaction in the contact center from a first party about an issue;
retrieving information from a record associated with the first party, wherein the information comprises at least one of:
previous orders,
total value of the orders,
cost of goods sold,
cost to return an item,
interaction history of the first party with the contact center,
end-call codes from each interaction in the history of the first party,
sentiment analysis of historical interaction transcripts, and
manually input information from an agent of the contact center;
analyzing the chat interaction for named entity recognition, topic identification, and sentiment analysis for each chat interaction of the first party to automatically determine the issue behind the interaction and determining a primary emotion among a plurality of emotions of the first party from the interaction by classifying the sentiment analysis using a neural network with LSTM units;
matching an action to the emotion of the first party where each action is associated with an intent, and the interaction is represented by a series of data points associated with intent of the interaction;
predicting customer disposition and reception to a course of resolution using the series of data points associated with the intent to resolve the issue through a plurality of means, considering a specified business priority comprising one of reducing average handling time or increasing a rate of first call resolution, wherein those means comprise at least one of:
troubleshooting,
refund,
return, and
reimbursement; and
returning a suggested course of action to the agent that maximizes the business priority.
8. The method of claim 7 , wherein the method further comprises the step of recording data on whether a course of action followed by the agent followed the suggested course of action and inputting the recorded data into the system as a feedback loop for improving future suggested courses of action.
9. The method of claim 7 , wherein the determining comprises using an algorithm formulated based on sequence classification from contact center historical data.
10. The method of claim 9 , wherein the contact center historical data further comprises: product loss, labor cost, cost of escalations, cost of public relations, interaction cost, and potential loss of future sales.
11. The method of claim 7 , wherein inputs to the neural network comprise at least one of: sentiment analysis of the interaction, tracking flags, mood analytics, chat analytics, image-based motion detection, and end-interaction codes.
12. The method of claim 7 , wherein the suggested course of action comprises a rationale for the suggestion and an authorization for the agent to conduct said course of action.
13. A system for automatically providing recommendations to an agent in a contact center environment, the system comprising:
a processor; and
a memory in communication with the processor, the memory storing instructions that, when executed by the processor causes the processor to return a suggested course of action by:
receiving a voice interaction in the contact center from a first party about an issue;
retrieving information from a record associated with the first party, wherein the information comprises at least one of:
previous orders,
total value of the orders,
cost of goods sold,
cost to return an item,
interaction history of the first party with the contact center,
end-call codes from each interaction in the history of the first party,
sentiment analysis of historical interaction transcripts, and
manually input information from an agent of the contact center;
generating a transcription of the voice interaction in real-time through a speech recognition service;
analyzing the transcription for named entity recognition, topic identification, and sentiment analysis for each utterance of the first party to automatically determine the issue behind the interaction and determining a primary emotion among a plurality of emotions of the first party from the transcription by classifying the sentiment analysis using a neural network with LSTM units;
matching an action to the emotion of the first party where each action is associated with an intent, and the interaction is represented by a series of data points associated with intent of the interaction;
predicting customer disposition and reception to a course of resolution using the series of data points associated with the intent to resolve the issue through a plurality of means, considering a specified business priority comprising one of reducing average handling time or increasing a rate of first call resolution, wherein those means comprise at least one of:
troubleshooting,
refund,
return, and
reimbursement; and
returning a suggested course of action to the agent that maximizes the business priority.
14. The system of claim 13 , wherein the memory in communication with the processor further causes the processor to return a suggested course of action by recording data on whether a course of action followed by the agent followed the suggested course of action and inputting the recorded data into the system as a feedback loop for improving future suggested courses of action.
15. The system of claim 13 , wherein the determining comprises using an algorithm formulated based on sequence classification from contact center historical data, and wherein the contact center historical data further comprises at least one of: product loss, labor cost, cost of escalations, cost of public relations, interaction cost, and potential loss of future sales.
16. The system of claim 13 , wherein inputs to the neural network comprise at least one of: sentiment analysis of the transcription, call tracking flags, mood analytics, speech analytics, image-based motion detection, and end-call codes.
17. The system of claim 13 , wherein the suggested course of action comprises a rationale for the suggestion and an authorization for the agent to conduct said course of action.