US2019312827A1PendingUtilityA1

Passing emotional chatbot sessions to the best suited agent

Assignee: IBMPriority: Aug 8, 2017Filed: Jun 25, 2019Published: Oct 10, 2019
Est. expiryAug 8, 2037(~11 yrs left)· nominal 20-yr term from priority
G06F 40/30H04L 51/04G06Q 30/016H04L 51/02G06Q 10/04G06F 17/2785
52
PatentIndex Score
0
Cited by
0
References
0
Claims

Abstract

Embodiments of the present invention disclose a method for an automated chat bot conversation session and an agent transfer system for the conversation session. The computer receives a user input from a user in an automated chat bot conversation session. The computer analyzes the user input for at least one sentiment, wherein an at least one analysis result is a value assigned to the at least one sentiment contained within the user input. The computer compares the at least one analysis result to a threshold value to determine if the user should be transferred from the automated chat bot conversation session to a conversation session with a suitable agent. The computer then transfers the user to the conversation session with the suitable agent.

Claims

exact text as granted — not AI-modified
What is claimed is: 
     
         1 . A method for an automated chat bot conversation session and an agent transfer system for the conversation session, the method comprises:
 receiving, by a computer, a user input from a user in an automated chat bot conversation session;   analyzing, by the computer, the user input for at least one sentiment contained within the user input, wherein an at least one analysis result is a value assigned to the at least one sentiment;   comparing, by the computer, the at least one analysis result to a threshold value to determine when to transfer the user from the automated chat bot conversation session to a conversation session with a suitable agent; and   transferring, by the computer, the user to the conversation session with the suitable agent, when the value of the at least one analysis result exceeds the threshold value.   
     
     
         2 . The method of  claim 1 , wherein the comparing, by the computer, the at least one analysis result to the threshold value, further comprises:
 increasing, by the computer, a counter by one for each of a plurality of sentiments that exceed the threshold value for the at least one sentiment, respectively;   comparing, by the computer, the number on the counter to a cumulative threshold value; and   in response to the number of the counter exceeding the cumulative threshold value, transferring, by the computer, the user to the conversation session with the suitable agent.   
     
     
         3 . The method of  claim 1 , wherein the at least one analysis result further comprises:
 assigning, by the computer, a band number for the at least one sentiment, wherein the band number comprises a range of values assigned to the at least one sentiment; and   assigning, by the computer, a switching score to the band number for the at least one sentiment.   
     
     
         4 . The method of  claim 3 , further comprises:
 determining, by the computer, a sum of the switching scores by adding the switching score for each of the at least one sentiment in the user input; and   transferring, by the computer, the user to the conversation session with the suitable agent when the sum of the switching score exceeds a switching score threshold that triggers the transfer to the conversation session with the suitable agent, wherein the threshold value is the switching score threshold.   
     
     
         5 . The method of  claim 1 , further comprises:
 determining, by the computer, a chat topic for the conversation session; and   retrieving, by the computer, at least one similar historic chat transcripts from a database, wherein the similar historic transcripts is based on the at least one analysis result and the determined chat topic.   
     
     
         6 . The method of  claim 5 , wherein the analyzing, by the computer, the user input, further comprises:
 creating, by the computer, a sentiment vector in sentiment vector space for the at least one sentiment contained within the user input;   assigning, by the computer, a switching score to the sentiment vector for the at least one sentiment contained within the user input by averaging a switching score for each of a plurality of the nearest neighbor historic chat sentiment vectors in sentiment vector space; and   determining, by the computer, whether the averaged switching score exceeds the threshold value that triggers the transfer to the conversation session with the suitable agent.   
     
     
         7 . The method of  claim 5 , further comprises:
 determining, by the computer, a plurality of suitable agents to respond to the user input, wherein each of the plurality of the suitable agents is skilled in the determined chat topic and has provided a positive outcome in a plurality of similar conversation sessions based on the retrieved historic chats; and   ranking, by the computer, the plurality of suitable agents based on provided a positive outcome in a plurality of similar conversations based on the retrieved historic chats.   
     
     
         8 . The method of  claim 7 , further comprises:
 determining, by the computer, availability of the plurality of suitable agents;   transmitting, by the computer, to the user the availability of the plurality of suitable agents and their ranking; and   receiving, by the computer, a user selection of the suitable agent from the transmitted availability of the plurality of suitable agents and their ranking;   wherein the user is transferred to the selected suitable agent.   
     
     
         9 . The method of  claim 1 , further comprises:
 determining, by the computer, whether the conversation session with the suitable agent is complete;   in response to determining that the conversation session is complete, soliciting, by the computer, a feedback from the user and the suitable agent that the user conducted the conversation session with;   receiving, by the computer, the feedback from the user and the suitable agent that the user conducted the conversation session with, wherein, the feedback reflects a plurality of opinions of the user and the suitable agent with regard to their satisfaction with the outcome of the conversation session; and   storing, by the computer, an identification number of the suitable agent, a conversation transcript based on the conversation between the user and the suitable agent, the user feedback, and the suitable agent feedback in the chat history database.   
     
     
         10 . The method of  claim 1 , wherein the receiving user input further comprises receiving a plurality of user inputs, and wherein analyzing, by the computer, the user input for at least one sentiment, further comprise analyzing the each of the plurality of user inputs for at least one sentiment, wherein an at least one analysis result is a value assigned to the at least one sentiment contained within each of the plurality of user inputs. 
     
     
         11 . The method of  claim 10 , wherein the comparing, by the computer, the at least one analysis result to the threshold value, further comprises:
 increasing, by the computer, a counter by one for a plurality of sentiments that exceed the threshold value for the at least one sentiment;   comparing, by the computer, the number on the counter to a cumulative threshold value; and   in response to the number of the counter exceeding the cumulative threshold value, transferring, by the computer, the user to the conversation session with the suitable agent.   
     
     
         12 . The method of  claim 10 , wherein analyzing the plurality of user inputs further comprises:
 assigning, by the computer, a band number for each of the at least one sentiment in each of the plurality of user inputs, wherein the band number comprises a range of values assigned to the at least one sentiment;   assigning, by the computer, a switching score to the band number for each of the at least one sentiment in each of the plurality of user inputs;   determining, by the computer, a sum of the switching score by adding the switching score for each of the at least one sentiment in each of the plurality of user inputs; and   transferring, by the computer, the user to the conversation session with the suitable agent when the sum of the switching score from the plurality of user inputs exceeds the threshold value that triggers the transfer to the conversation session with the suitable agent.   
     
     
         13 . The method of  claim 10 , further comprises:
 determining, by the computer, a chat topic for the conversation session;   retrieving, by the computer, at least one similar historic chat transcripts from a database, wherein the similar historic transcripts is based on the at least one analysis result and the determined chat topic;   wherein analyzing the plurality of user inputs further comprises;   creating, by the computer, a sentiment vector in sentiment vector space corresponding to each of the plurality of user inputs;   assigning, by the computer, a switching score to the sentiment vector of each of the plurality of user inputs for the at least one sentiment contained within the user input by averaging a switching score for a plurality of the nearest neighbor historic chat sentiment vectors in sentiment vector space;   determining, by the computer, a sum of the switching scores by adding the switching score for each of the sentiment vectors corresponding to the plurality of user inputs; and   determining, by the computer, whether the sum of the averaged switching scores exceeds the threshold value that triggers the transfer to the conversation session with the suitable agent.

Join the waitlist — get patent alerts

Track US2019312827A1 — get alerts on status changes and closely related new filings.

We store only your email — no account needed. See our privacy policy.