The Web as a Knowledge-base for Answering Complex Questions
A. Talmor, и J. Berant. (2018)cite arxiv:1803.06643Comment: accepted as a long paper at NAACL 2018.
Аннотация
Answering complex questions is a time-consuming activity for humans that
requires reasoning and integration of information. Recent work on reading
comprehension made headway in answering simple questions, but tackling complex
questions is still an ongoing research challenge. Conversely, semantic parsers
have been successful at handling compositionality, but only when the
information resides in a target knowledge-base. In this paper, we present a
novel framework for answering broad and complex questions, assuming answering
simple questions is possible using a search engine and a reading comprehension
model. We propose to decompose complex questions into a sequence of simple
questions, and compute the final answer from the sequence of answers. To
illustrate the viability of our approach, we create a new dataset of complex
questions, ComplexWebQuestions, and present a model that decomposes questions
and interacts with the web to compute an answer. We empirically demonstrate
that question decomposition improves performance from 20.8 precision@1 to 27.5
precision@1 on this new dataset.
Описание
[1803.06643] The Web as a Knowledge-base for Answering Complex Questions
%0 Generic
%1 talmor2018knowledgebase
%A Talmor, Alon
%A Berant, Jonathan
%D 2018
%K bert_performance dataset instituteclustering knwoledge web
%T The Web as a Knowledge-base for Answering Complex Questions
%U http://arxiv.org/abs/1803.06643
%X Answering complex questions is a time-consuming activity for humans that
requires reasoning and integration of information. Recent work on reading
comprehension made headway in answering simple questions, but tackling complex
questions is still an ongoing research challenge. Conversely, semantic parsers
have been successful at handling compositionality, but only when the
information resides in a target knowledge-base. In this paper, we present a
novel framework for answering broad and complex questions, assuming answering
simple questions is possible using a search engine and a reading comprehension
model. We propose to decompose complex questions into a sequence of simple
questions, and compute the final answer from the sequence of answers. To
illustrate the viability of our approach, we create a new dataset of complex
questions, ComplexWebQuestions, and present a model that decomposes questions
and interacts with the web to compute an answer. We empirically demonstrate
that question decomposition improves performance from 20.8 precision@1 to 27.5
precision@1 on this new dataset.
@misc{talmor2018knowledgebase,
abstract = {Answering complex questions is a time-consuming activity for humans that
requires reasoning and integration of information. Recent work on reading
comprehension made headway in answering simple questions, but tackling complex
questions is still an ongoing research challenge. Conversely, semantic parsers
have been successful at handling compositionality, but only when the
information resides in a target knowledge-base. In this paper, we present a
novel framework for answering broad and complex questions, assuming answering
simple questions is possible using a search engine and a reading comprehension
model. We propose to decompose complex questions into a sequence of simple
questions, and compute the final answer from the sequence of answers. To
illustrate the viability of our approach, we create a new dataset of complex
questions, ComplexWebQuestions, and present a model that decomposes questions
and interacts with the web to compute an answer. We empirically demonstrate
that question decomposition improves performance from 20.8 precision@1 to 27.5
precision@1 on this new dataset.},
added-at = {2021-01-20T11:48:27.000+0100},
author = {Talmor, Alon and Berant, Jonathan},
biburl = {https://www.bibsonomy.org/bibtex/2c06461f7de2685f326d248d9e24d7488/parismic},
description = {[1803.06643] The Web as a Knowledge-base for Answering Complex Questions},
interhash = {972fa193640bbba9272ebdb24ed56673},
intrahash = {c06461f7de2685f326d248d9e24d7488},
keywords = {bert_performance dataset instituteclustering knwoledge web},
note = {cite arxiv:1803.06643Comment: accepted as a long paper at NAACL 2018},
timestamp = {2021-01-20T11:48:27.000+0100},
title = {The Web as a Knowledge-base for Answering Complex Questions},
url = {http://arxiv.org/abs/1803.06643},
year = 2018
}