Research on feature-based opinion mining using topic maps
| Pages | 435-456 |
| Date | 06 June 2016 |
| Published date | 06 June 2016 |
| DOI | https://doi.org/10.1108/EL-11-2014-0197 |
| Author | Lixin Xia,Zhongyi Wang,Chen Chen,Shanshan Zhai |
| Subject Matter | Information & knowledge management,Information & communications technology,Internet |
Research on feature-based
opinion mining using topic maps
Lixin Xia, Zhongyi Wang, Chen Chen and Shanshan Zhai
School of Information Management, Central China Normal University,
Wuhan City, People’s Republic of China
Abstract
Purpose – Opinion mining (OM), also known as “sentiment classication”, which aims to discover
common patterns of user opinions from their textual statements automatically or semi-automatically, is
not only useful for customers, but also for manufacturers. However, because of the complexity of
natural language, there are still some problems, such as domain dependence of sentiment words,
extraction of implicit features and others. The purpose of this paper is to propose an OM method based
on topic maps to solve these problems.
Design/methodology/approach – Domain-specic knowledge is key to solve problems in
feature-based OM. On the one hand, topic maps, as an ontology framework, are composed of topics,
associations, occurrences and scopes, and can represent a class of knowledge representation schemes.
On the other hand, compared with ontology, topic maps have many advantages. Thus, it is better to
integrate domain-specic knowledge into OM based on topic maps. This method can make full use of
the semantic relationships among feature words and sentiment words.
Findings – In feature-level OM, most of the existing research associate product features and opinions
by their explicit co-occurrence, or use syntax parsing to judge the modication relationship between
opinion words and product features within a review unit. They are mostly based on the structure of
language units without considering domain knowledge. Only few methods based on ontology
incorporate domain knowledge into feature-based OM, but they only use the “is-a” relation between
concepts. Therefore, this paper proposes feature-based OM using topic maps. The experimental results
revealed that this method can improve the accuracy of the OM. The ndings of this study not only
advance the state of OM research but also shed light on future research directions.
Research limitations/implications – To demonstrate the “feature-based OM using topic maps”
applications, this work implements a prototype that helps users to nd their new washing machines.
Originality/value – This paper presents a new method of feature-based OM using topic maps, which
can integrate domain-specic knowledge into feature-based OM effectively. This method can improve
the accuracy of the OM greatly. The proposed method can be applied across various application
domains, such as e-commerce and e-government.
Keywords Feature extraction, Sentiment classication, Topic map, Feature-based opinion mining
Paper type Research paper
Introduction
With the development of Web 2.0, which emphasizes the participation of users, the
number of online opinion sources is growing rapidly. More and more websites, such as
This study is supported by National Social Science Foundation of China: “Research on
Multi-granularity Integration Knowledge Services of Digital Library Based on Linked Data”
(14CTQ003) and is the major project of National Social Science Foundation of China: “Research on
Knowledge Discovery of Internet Resource base on Multi-dimensional Aggregation”
(No. 13&ZD183).
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0264-0473.htm
Research on
feature-based
opinion
mining
435
Received 13 November 2014
Revised 26 February 2015
27 April 2015
Accepted 10 June 2015
TheElectronic Library
Vol.34 No. 3, 2016
pp.435-456
©Emerald Group Publishing Limited
0264-0473
DOI 10.1108/EL-11-2014-0197
Amazon (www.amazon.com) and Epinions (www.epinions.com), encourage users to
write opinions for the products they are interested in. These online opinions are useful
for customers and manufacturers. The customers refer to these opinions to help decide
about their purchases and manufacturers can gather feedback from their customers to
improve the quality of their products. However, these online opinions are text
information generated by a large number of users, which results in unstructured and
unmanaged content. The opinions can number in the hundreds or even in the thousands,
so it may be difcult for people to use them. Therefore, automatic extraction and
summarization of these opinions has become an urgent need. To meet this requirement,
the technology of opinion mining (OM – also known as “sentiment classication”),
which aims to discover common patterns of user opinions from their textual statements
automatically or semi-automatically, initially proposed by Hatzivassiloglou and
McKeown (1997), has now become a signicant area of research in the eld of data
mining. The method involves techniques from different disciplines, including
information retrieval, natural language processing and data mining (Vijaya and Sudha,
2013). Meanwhile, a considerable number of varied research results (Choi et al., 2005;
Dave et al., 2003;Ghose et al., 2007;Hu and Liu, 2004a,2004b;Liu, 2010;Pang and Lee,
2008;Turney, 2002) have been achieved, which can be mainly divided into three
directions: document-level OM, sentence-level OM and feature-level OM.
For document-level OM (Pang et al., 2002), the entire document is classied as
positive, negative or neutral. However, this method is too broad. In most cases, both
positive and negative opinions can appear in the same document, so the sentiment
orientation at the document level is not sufcient. At the sentence level (Wiebe and
Riloff, 2005;Wiebe et al., 2004;Wilson et al., 2005), sentiment classication is applied to
individual sentences in a document. Although OM at the sentence level is useful in many
cases, it still leaves much to be desired. A positive evaluative sentence on a particular
entity does not mean that the author has positive opinions on every aspect of the entity.
Researchers, such as Hu and Liu (2004a,2004b) and Liu et al. (2005), have worked on
ner-grained OM which predicts the sentiment orientation related to different review
features. The task is known as feature-level OM. In feature-level OM, structured
opinions on individual features of a whole object are extracted from subjective texts, so
it is important to determine the reviewers’ opinions towards different product features
instead of the overall opinion in those reviews. For this paper, the authors focus on
feature-based OM. The authors propose an OM approach and develop an OM system
based on the proposed approach to conduct sentiment analysis on the reviews of
washing machines. The authors also provide an interface for showing summaries of OM
results and report the results. Conclusions and future work are presented.
Literature review
Feature-based OM has been studied by many researchers in recent years. The rst
denition of the problem can be found in Hu and Liu (2004a):
[…] given a set of customer reviews of a particular product, the task involves three subtasks:
identifying features of the product; for each feature, identifying review sentences; and
producing a summary using the discovered information.
Many different kinds of approaches for feature-based OM have been proposed, such as
Abbasi (2003),Baccianella et al. (2010),Ding and Liu (2007) and Zhu et al. (2011).
EL
34,3
436
To continue reading
Request your trial