Browse Prior Art Database

Returning matching passage as abstract in hit-list

IP.com Disclosure Number: IPCOM000013605D
Original Publication Date: 2000-Mar-01
Included in the Prior Art Database: 2003-Jun-18
Document File: 1 page(s) / 37K

Publishing Venue

IBM

Abstract

Disclosed is a system and method for providing users with more the information they are seeking by returning matching passages as abstracts in hit-lists. Traditional search engines return to the user a list of the top-matching documents. Sometimes the document title is accompanied by what is called a document abstract, but it just amounts to the first 200 or so characters of the document. It is often the case that not only does the title+abstract combi- nation not answer the user's query, but it is still unclear whether reading the document will elicit the answer. The user must link to the document, and then proceed to read some or all of the document in order to discover if the answer is really there. In the context of a Question-Answering system such as the one described in the paper "The Use of Predictive Annotation for Question Answering in TREC8" in the Proceedings of the 8th Text Retrieval Conference, to be published by NIST in Spring '00, the system not only finds the best-matching documents, but specific passages within those documents that answer the query. In particular, each document in the hit-list will have a "best-passage" associated with it. Our invention is to show this best matching passage in lieu of the abstract. In this way, users often will often get their answers without having to link to and read the document itself. 1

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 100% of the total text.

Page 1 of 1

Returning matching passage as abstract in hit-list

    Disclosed is a system and method for providing users with more the information they are seeking by returning matching passages as abstracts in hit-lists. Traditional search engines return to the user a list of the top-matching documents. Sometimes the document title is accompanied by what is called a document abstract, but it just amounts to the first 200 or so characters of the document. It is often the case that not only does the title+abstract combi- nation not answer the user's query, but it is still unclear whether reading the document will elicit the answer. The user must link to the document, and then proceed to read some or all of the document in order to discover if the answer is really there.

    In the context of a Question-Answering system such as the one described in the paper "The Use of Predictive Annotation for Question Answering in TREC8" in the Proceedings of the 8th Text Retrieval Conference, to be published by NIST in Spring '00, the system not only finds the best-matching documents, but specific passages within those documents that answer the query. In particular, each document in the hit-list will have a "best-passage" associated with it. Our invention is to show this best matching passage in lieu of the abstract. In this way, users often will often get their answers without having to link to and read the document itself.

1