In this paper, we apply a general deep learning (DL) framework for the answer
selection task, which does not depend on manually defined features or
linguistic tools. The basic framework is to build the embeddings of questions
and answers based on bidirectional long short-term memory (biLSTM) models, and
measure their closeness by cosine similarity. We further extend this basic
model in two directions. One direction is to define a more composite
representation for questions and answers by combining convolutional neural
network with the basic framework. The other direction is to utilize a simple
but efficient attention mechanism in order to generate the answer
representation according to the question context. Several variations of models
are provided. The models are examined by two datasets, including TREC-QA and
InsuranceQA. Experimental results demonstrate that the proposed models
substantially outperform several strong baselines.
Description
[1511.04108] LSTM-based Deep Learning Models for Non-factoid Answer Selection
%0 Generic
%1 tan2015lstmbased
%A Tan, Ming
%A Santos, Cicero dos
%A Xiang, Bing
%A Zhou, Bowen
%D 2015
%K answer dl lstm selection
%T LSTM-based Deep Learning Models for Non-factoid Answer Selection
%U http://arxiv.org/abs/1511.04108
%X In this paper, we apply a general deep learning (DL) framework for the answer
selection task, which does not depend on manually defined features or
linguistic tools. The basic framework is to build the embeddings of questions
and answers based on bidirectional long short-term memory (biLSTM) models, and
measure their closeness by cosine similarity. We further extend this basic
model in two directions. One direction is to define a more composite
representation for questions and answers by combining convolutional neural
network with the basic framework. The other direction is to utilize a simple
but efficient attention mechanism in order to generate the answer
representation according to the question context. Several variations of models
are provided. The models are examined by two datasets, including TREC-QA and
InsuranceQA. Experimental results demonstrate that the proposed models
substantially outperform several strong baselines.
@misc{tan2015lstmbased,
abstract = {In this paper, we apply a general deep learning (DL) framework for the answer
selection task, which does not depend on manually defined features or
linguistic tools. The basic framework is to build the embeddings of questions
and answers based on bidirectional long short-term memory (biLSTM) models, and
measure their closeness by cosine similarity. We further extend this basic
model in two directions. One direction is to define a more composite
representation for questions and answers by combining convolutional neural
network with the basic framework. The other direction is to utilize a simple
but efficient attention mechanism in order to generate the answer
representation according to the question context. Several variations of models
are provided. The models are examined by two datasets, including TREC-QA and
InsuranceQA. Experimental results demonstrate that the proposed models
substantially outperform several strong baselines.},
added-at = {2017-08-16T15:06:03.000+0200},
author = {Tan, Ming and Santos, Cicero dos and Xiang, Bing and Zhou, Bowen},
biburl = {https://www.bibsonomy.org/bibtex/278c8c1c1e50daf7588b4c04c41b4294e/nosebrain},
description = {[1511.04108] LSTM-based Deep Learning Models for Non-factoid Answer Selection},
interhash = {a24b638932677833b739a1d74bcb7e8e},
intrahash = {78c8c1c1e50daf7588b4c04c41b4294e},
keywords = {answer dl lstm selection},
note = {cite arxiv:1511.04108Comment: added new experiments on TREC-QA},
timestamp = {2017-08-16T15:06:03.000+0200},
title = {LSTM-based Deep Learning Models for Non-factoid Answer Selection},
url = {http://arxiv.org/abs/1511.04108},
year = 2015
}