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

System and method for user defined and dynamic control of confidence versus performance in complex analytical processing pipelines

IP.com Disclosure Number: IPCOM000235640D
Publication Date: 2014-Mar-17
Document File: 2 page(s) / 26K

Publishing Venue

The IP.com Prior Art Database

Abstract

A system and method for user defined and dynamic control of confidence versus performance in complex analytical processing pipelines is disclosed. User options include: 1) Choose their desired minimum confidence level before analyzing their case. 2) Understand the tradeoffs between confidence and response time. 3) Dynamically adapt the analytics and associated computational requirements to fit #1 and #2.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.

Page 01 of 2

System and method for user defined and dynamic control of confidence versus performance in complex analytical processing pipelines

Disclosed is a system and method for user defined and dynamic control of confidence versus performance in complex analytical processing pipelines.

IBM Watson

          has ushered in a new class of cognitive applications that use advanced analytic techniques to assist a human in making better, evidence based decisions by deep natural language analysis of all case relevant information. In this style of application, the end user is not necessarily looking for the one right answer, but potentially a list of best possible answers - supported by probabilistic confidence values

and evidence. Current approaches to providing this capability (e.g. JEOPARDY!

® IBM

Watson) attempt to provide the highest degree of confidence possible for every outcome - resulting in the highest computational overhead and longest response time for every scenario. Not every application or end user requires this level of confidence and in fact some might even prefer faster response time at the expense of confidence - current systems do not allow an end user to:

Choose their desired minimum confidence level before analyzing their case


1.

Understand the tradeoffs between confidence and response time.


2.

Dynamically adapt the analytics and associated computational requirements to fit #1


3.

and #2.

There are no known solutions that allow dynamic control. It is possible to statically define different paths through the pipeline analytics or deploy different pipelines that would offer varying degrees of performance and accuracy, but these would be hard coded solutions.

The disclosed approach revolves around four things:

User interface controls (slider, dial, option chooser, etc.) that allow the end user to


1.

select a desired level of machine confidence i...