Browse Prior Art Database

Improving User Feedback In A Question Answering System For Indirect Answers

IP.com Disclosure Number: IPCOM000239021D
Publication Date: 2014-Oct-01
Document File: 3 page(s) / 39K

Publishing Venue

The IP.com Prior Art Database

Abstract

A system and method for improving user feedback in a question answering system for partial or indirect answers is disclosed. An indication of incomplete but relevant is returned to the user.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 34% of the total text.

Page 01 of 3

Improving User Feedback In A Question Answering System For Indirect Answers

Disclosed is a system and method for improving user feedback in a question answering system for partial or indirect answers. An indication of incomplete but relevant is returned to the user.

In some cases, only a partial answer to a question or mostly, but not completely correct answer (an "indirect" or "ballpark" answer) can be determined by a question answering system. A question answering system may not always have a precise answer to a given question for any number of reasons. However, it may find answers that are relevant to the topic even though they do not directly answer the question. Usually indirect answers are returned to the user in the same way a correct answer is returned. Indirect answers may be perceived as incorrect even if they contain useful information that is related to the question. This experience may give the impression that the system is unintelligent and didn't understand the question. Disclosed is a method in which the system behaves in a more cognitively aware manner, by explicitly stating that is doesn't have a precise answer, but that it has information that may be relevant. This system gives more natural responses from a regular conversational standpoint and better interaction. This kind of response signals that the system actually understood the question and understands that the answers it retrieved are not precise, but still topical. This behavior goes a long way to improve the user experience and promote the perception of the system as cognitively aware and intelligent.

Question answering systems are not always able to answer a given question precisely. This could be due to a variety of reasons ranging from system performance to missing information in the knowledge resources. Current question answering systems find likely answers and return them to the user, even if they are not a direct answer to the question. Answers that address the general topic but do not directly answer the question, i.e. indirect or "ballpark" answers, are typically perceived as wrong, even if they supply related information. In addition, these answers can be misleading if they are perceived as correct. However, these answers may still be useful to the user. So the system is faced with two bad alternatives:

Present an answer to the user that might be useful but is probably wrong, and risk


1.

misleading the user or being perceived as unintelligent.

Present no answer to the user, and miss an opportunity to provide information that


2.

might be useful.

The disclosed method presents a third choice: The system explicitly acknowledges that it does not have an answer, that it only has information that may be relevant, and presents the user with that answer only after explicitly acknowledging its limitations. With this approach. the user gets potentially useful information related to their question.

Another advantage of this kind of system response is that...