Browse Prior Art Database

System and method to improve accuracy, confidence, evidence, relevant and expected answers based on the context of the question asker

IP.com Disclosure Number: IPCOM000238560D
Publication Date: 2014-Sep-03
Document File: 2 page(s) / 31K

Publishing Venue

The IP.com Prior Art Database

Abstract

A system and method for determining how the context of the asker can be used to provide relevant answers to a question based on information about the asker is disclosed.

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 to improve accuracy , confidence, evidence, relevant and expected answers based on the context of the question asker

Disclosed is a system and method for determining how the context of the asker can be used to provide relevant answers to a question based on information about the asker. Social as well as other sources are examined as a source of information and how the question can be parsed.

A correct answer to a question in a question-and-answering system is often dependent on who is asking the question. As an example, the question "Who is the greatest football player who ever lived?" very much depends on the context of the asker as being either North American or European influenced.

Increasing accuracy and providing an expected answer can be accomplished by knowing facts and context about the asker and taking them into account when providing answers and evidence. Context of the user refers to:

Their profession and affiliations


Their location
Their lineage and background
Their likes and interests
Groups they are associated with
Types of questions and answers received in the past

The source of information that the question answering system draws on to answer the question should be determined by that context. The context may be stored as part of a user profile.

The user context can also be enhanced based on the following:
Previous questions the user has asked and what Corpora they went against. i.e. for the last 20 questions from user A, Corpus X and Corpus Y were accessed. User A is associated with a group of Users. Using a clustering algorithm (or other learning algorithm), the system knows that users of this group access Corpora X and Y 95%, therefore User A will only use Corpora X and Y. Feedback (i.e. what passages the user clicks on and tracing back to the corpora they originated from)

User A likes Football and lives in location L. The system knows that from other users similar to A who like Football and Live near location L rate Corpora X and
Y highest, but not Z, therefore A's question will go against Corpora X and Y.

...