System and method for a cognitive system plug-in answering subject matter expert questions

Abstract

Embodiments provide a system and method for integrating a cognitive system into a call center. The system and method include ingesting, in real-time through an instant messaging application, one or more questions from one or more call center agents; ingesting, in real-time through the instant messaging application, one or more answers associated with the one or more questions; storing one or more question and answer pairs in a corpus; analyzing, through a cognitive system, the corpus of the one or more question and answer pairs; receiving, through the instant messaging system, one or more additional questions; determining a proposed answer to each additional question based on the analysis of the corpus; analyzing, through the cognitive system, the proposed answer; and incorporating the analysis of the proposed answer into the analysis of the one or more question and answer pairs. The answers can be provided by subject matter experts or call center managers. The cognitive system can interface with the instant messaging system through the use of a plug-in module, which can use one or more registration commands to facilitate the interface.

What is claimed is: 
     
         1 . A method of training a cognitive system in a working call center, the method comprising:
 receiving one or more questions collected through an instant messenger application;   receiving one or more answers collected through the instant messenger application,   wherein the one or more answers are associated with the one or more questions;   receiving an additional question collected through the instant messenger application;   determining a proposed answer to the additional question comprising:
 transforming the additional question into an acyclic graph to determine a knowledge domain of the additional question; 
 identifying element clusters associated with the one or more questions and the one or more answers that are compatible with the knowledge domain of the additional question; 
 ranking the identified element clusters; and 
 selecting the identified element clusters ranked highest for use in determining the proposed answer; and 
   analyzing the proposed answer to train the cognitive system.   
     
     
         2 . The method of  claim 1 , further comprising:
 incorporating the analysis of the proposed answer into an analysis of future questions.   
     
     
         3 . The method of  claim 2 , further comprising:
 wherein the analysis of the proposed answer determines a likelihood that the proposed answer is a correct answer to the additional question by performing a deep analysis using a plurality of reasoning algorithms.   
     
     
         4 . The method of  claim 1 , further comprising:
 wherein the analysis of the proposed answer comprises receiving an evaluation of the proposed answer from a subject matter expert; and   incorporating the evaluation of the proposed answer into an analysis of future questions.   
     
     
         5 . The method of  claim 1 , further comprising:
 defining an element data structure comprising one or more knowledge domains of the questions and the answers; and   analyzing the element data structure to identify the element clusters based on their characteristics.   
     
     
         6 . The method of  claim 1 , further comprising:
 outputting the proposed answer through the instant messenger application when the cognitive system has permission to provide the proposed answer.   
     
     
         7 . The method of  claim 6 , further comprising:
 storing similar element clusters in one or more cluster data structures.   
     
     
         8 . The method of  claim 7 , further comprising:
 searching the one or more cluster data structures to identify the element clusters associated with the one or more questions and the one or more answers that are compatible with the knowledge domain of the additional question.   
     
     
         9 . The method of  claim 1 , further comprising:
 wherein the acyclic graphs comprise nodes representing potential facts and connectors representing the overall connections between the potential facts.   
     
     
         10 . The method of  claim 3 , further comprising:
 wherein each reasoning algorithm performs a separate type of analysis of the language of the additional question language and the language of the proposed answer.   
     
     
         11 . A call center question and answer system comprising:
 an instant messaging application, wherein the instant messaging application has one or more chat channels;   a plug-in module, wherein the plug-in module is configured to interface with the instant messaging application; and   a cognitive system integrated into the call center question and answer system, wherein the cognitive system is configured to receive data from the instant messaging application and perform a deep analysis and comparison of one or more input questions and one or more candidate answers.   
     
     
         12 . The system as recited in  claim 11 , further comprising:
 wherein the plug-in module is configured to interface with the instant messaging application in real-time.   
     
     
         13 . The system as recited in  claim 11 , further comprising:
 wherein the plug-in module is configured to collect data transmitted through the instant messaging application.   
     
     
         14 . The system as recited in  claim 11 , further comprising:
 wherein the plug-in module is configured to register itself with the instant messaging application using the one or more registration commands.   
     
     
         15 . The system as recited in  claim 11 , further comprising:
 wherein the plug-in module is configured to collect data using one or more registration commands.   
     
     
         16 . The system as recited in  claim 11 , further comprising:
 wherein the cognitive system is configured to do deep analysis using a plurality of reasoning algorithms, each performing a separate type of analysis of the language of the one or more input questions and the one or more candidate answers.   
     
     
         17 . The system as recited in  claim 16 , further comprising:
 wherein each reasoning algorithm generates a score based on the analysis performed by the reasoning algorithm, wherein the score is a measure of correctness for the one or more candidate answers.   
     
     
         18 . The system as recited in  claim 17 , further comprising:
 wherein the cognitive system is configured to synthesize the score from each reasoning algorithm into a confidence score for each candidate answer.   
     
     
         19 . The system as recited in  claim 18 , further comprising:
 wherein the cognitive system is configured to rank the confidence scores and compare them to predetermined thresholds.   
     
     
         20 . A computer implemented method, in a data processing system comprising a processor and a memory comprising instructions which are executed by the processor to cause the processor to implement training for a cognitive system in a working call center, the method comprising:
 receiving one or more answers collected through the instant messenger application, wherein the one or more answers are associated with the one or more questions;   receiving an additional question collected through the instant messenger application; providing a proposed answer to the additional question comprising:
 transforming the additional question into an acyclic graph to determine a knowledge domain of the additional question; 
 identifying element clusters associated with the one or more questions and the one or more answers that are compatible with the knowledge domain of the additional question; 
 ranking the identified element clusters; 
 selecting the identified element clusters ranked highest for use in determining the proposed answer; and 
   analyzing the proposed answer to train the cognitive system.   analyzing the proposed answer, wherein the analysis of the proposed answer determines a likelihood that the proposed answer is a correct answer to the additional question by performing a deep analysis using a plurality of reasoning algorithms; and incorporating the analysis of the proposed answer into an analysis of future questions.