Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Expert Systems coordinating to answer a Complex Question

IP.com Disclosure Number: IPCOM000234127D
Publication Date: 2014-Jan-13
Document File: 2 page(s) / 59K

Publishing Venue

The IP.com Prior Art Database

Abstract

System and method for expert systems to coordinate to answer a complex question

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 2

Expert Systems coordinating to answer a Complex Question

Background


In some situations there can be a complex question that requires several expert question-and-answering (Q/A) systems, each with specific expertise in their own domain, to coordinate to provide an accurate answer.

For example:


A middleware vendor may have an expert Q/A system that answers questions on an application server.

A database vendor may have an expert Q/A system that answers questions on a database.

A hardware vendor may have an expert Q/A system that answers question on the hardware setup and underlying OS.

These may form a solution in a customer's environment. Troubleshooting problems in this kind of environment generally requires coordination from representatives of each company. In the proposed system, the expert systems above will coordinate to answer troubleshooting questions about the overall solution.

The system may be augmented with logs, but primarily the system uses corpora ingested for the domain and application, in addition to training, to be able to answer questions

Solution overview


Disclosed is a method to have each expert Q/A system acknowledge the expertise of each other system and formulate questions for another system to help make a decision on an original question, that could not be answered accurately by the initial or main expert Q/A system.

The question is submitted to an expert system with context and supporting information for any unanswered or ambiguous statements or questions produced another expert system is asked a clarifying question.

Solution details


The following steps outline the solution


1) The user asks a question to an expertise system. (Question may be augmented with logs and/or structured data)


2) The expertise system finds an initial set of candidate answers with supporting evidence. (The system can also denote "missing" evidence)


3) The expertise system finds the n-grams and keywords from the answer and supporting/missing evidence


4) These n-grams and keywords are sent to a "Registry of Expertise Systems" to find expertise systems that could supply additional insight to the question a) Expertise systems will register to this registry for n top keywords/n-grams from their corpus, for instance n=1000

b) A simple matching algorithm is matching n-gram similarity, if the match score is over a certain threshold that system will be asked a question


5) The expertise syst...