Reviews of products or services on Internet marketplace websites contain a
rich amount of information. Users often wish to survey reviews or review
snippets from the perspective of a certain aspect, which has resulted in a
large body of work on aspect identification and extraction from such corpora.
In this work, we evaluate a newly-proposed neural model for aspect extraction
on two practical tasks. The first is to extract canonical sentences of various
aspects from reviews, and is judged by human evaluators against alternatives. A
$k$-means baseline does remarkably well in this setting. The second experiment
focuses on the suitability of the recovered aspect distributions to represent
users by the reviews they have written. Through a set of review reranking
experiments, we find that aspect-based profiles can largely capture notions of
user preferences, by showing that divergent users generate markedly different
review rankings.
Description
- Use ABAE on airbnb reviews
- Attempt to extract canonical sentences from review for summarization
- Attempt to create user profiles from their reviews that reveals their preferences for personalization
- Work by some undergrads, not super awesome, but may have some interesting ideas
%0 Report
%1 mitcheltree2018using
%A Mitcheltree, Christopher
%A Wharton, Veronica
%A Saluja, Avneesh
%D 2018
%K abae airbnb deeplearning reviews
%T Using Aspect Extraction Approaches to Generate Review Summaries and User
Profiles
%U http://arxiv.org/abs/1804.08666
%X Reviews of products or services on Internet marketplace websites contain a
rich amount of information. Users often wish to survey reviews or review
snippets from the perspective of a certain aspect, which has resulted in a
large body of work on aspect identification and extraction from such corpora.
In this work, we evaluate a newly-proposed neural model for aspect extraction
on two practical tasks. The first is to extract canonical sentences of various
aspects from reviews, and is judged by human evaluators against alternatives. A
$k$-means baseline does remarkably well in this setting. The second experiment
focuses on the suitability of the recovered aspect distributions to represent
users by the reviews they have written. Through a set of review reranking
experiments, we find that aspect-based profiles can largely capture notions of
user preferences, by showing that divergent users generate markedly different
review rankings.
@techreport{mitcheltree2018using,
abstract = {Reviews of products or services on Internet marketplace websites contain a
rich amount of information. Users often wish to survey reviews or review
snippets from the perspective of a certain aspect, which has resulted in a
large body of work on aspect identification and extraction from such corpora.
In this work, we evaluate a newly-proposed neural model for aspect extraction
on two practical tasks. The first is to extract canonical sentences of various
aspects from reviews, and is judged by human evaluators against alternatives. A
$k$-means baseline does remarkably well in this setting. The second experiment
focuses on the suitability of the recovered aspect distributions to represent
users by the reviews they have written. Through a set of review reranking
experiments, we find that aspect-based profiles can largely capture notions of
user preferences, by showing that divergent users generate markedly different
review rankings.},
added-at = {2019-04-11T19:54:10.000+0200},
author = {Mitcheltree, Christopher and Wharton, Veronica and Saluja, Avneesh},
biburl = {https://www.bibsonomy.org/bibtex/2a4b341c016c402094984a20161a40e15/vsathish},
description = {- Use ABAE on airbnb reviews
- Attempt to extract canonical sentences from review for summarization
- Attempt to create user profiles from their reviews that reveals their preferences for personalization
- Work by some undergrads, not super awesome, but may have some interesting ideas},
interhash = {b98c8b5277f438367c6c1078c7fd978b},
intrahash = {a4b341c016c402094984a20161a40e15},
keywords = {abae airbnb deeplearning reviews},
note = {cite arxiv:1804.08666Comment: Equal contribution from first two authors. Accepted for publication in the NAACL 2018 Industry Track},
timestamp = {2019-04-11T19:54:10.000+0200},
title = {Using Aspect Extraction Approaches to Generate Review Summaries and User
Profiles},
url = {http://arxiv.org/abs/1804.08666},
year = 2018
}