Browse Prior Art Database

Automated Forum using a Question Answering System

IP.com Disclosure Number: IPCOM000237219D
Publication Date: 2014-Jun-09
Document File: 2 page(s) / 29K

Publishing Venue

The IP.com Prior Art Database

Abstract

An automated forum using a question answering system is disclosed.

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

Page 01 of 2

Automated Forum using a Question Answering System

Disclosed is a system and method for automatically responding to questions posed in a forum using a Question Answer (QA) system powered by Natural Language Processing (NLP). The forum is used as the corpus for the QA system, which makes the corpus somewhat unique and different from existing corpora types.

Subject matter experts (SMEs) who regularly post to web forums are often inundated with similar questions over and over again. Forum posters are often lazy and do not take the time to properly search previous posts to find the answer to their questions. Those posters who do search for previous posts often have long lists of results to read through. SMEs often monitor all posts and respond to them. This takes a lot of time. Since time is limited, ideally questions should only need to be answered once. SMEs should be consulted when a question has not already been answered or needs to be further detailed.

With the disclosed method, appropriate SMEs are only consulted when new questions are posed that the QA system itself cannot answer. When the SME does answer, that answer becomes part of the QA system's knowledge base for future questions, thereby allowing the QA system to answer a previously answered question in the future.

The preferred embodiment of this method is a web hosted forum with the following steps:

A user posts a question in forum


1.

QA system monitors forum for new questions


2.

Using the forum as its corpus, the QA system attempts to answer the new question


3.

from the forum.

Forum corpus is treated as evidence with the following key additions:

Answers user ranked correct are weighted higher


a.

Answers voted high are adjusted for a higher weight


b.

Answers by SMEs for the domain and topic are associated higher


c.

QA system loads evidence with characteristics above, indexed against domain and SMEs posts
QA system identifies the Lexical Answer Type (LAT), domain and matches it against the evidence that was ingested as an answer

If QA system finds an answer with confidence above a given threshold:


a.

QA system responds to the forum post with the answer.

o

If QA system cannot find an answer with confidence above a given threshold:


b.

QA system identifies SME based on the users associated with previous

o

forum postings.

The QA system n...