New Trends in Opinion Mining Technologies
in the Industry
J.UCS Special Issue
Rafael Valencia-García (Informatics and Systems Department, Universidad de Murcia, Spain
valencia@um.es)
Ricardo Colomo-Palacios
(Computer Science Department, Østfold University College, Norway
ricardo.colomo-palacios@hiof.no)
Giner Alor-Hernández
(Division of Research and Postgraduate Studies, Instituto Tecnológico de Orizaba, Mexico
galor@itorizaba.edu.mx)
The Internet has become a global vehicle to express opinions and share
information. In this sense, the Web 2.0 phenomenon made the social
Web, initiating an explosion in the number of Web users and the amount
of available information. In this new scenario, collaboration with
customers has emerged through ordinary use of Web 2.0 from both
companies and clients alike. As a consequence of the endless
opportunities of the social use of the Web, customers today have
powerful tools to express their opinions and influence on business
systems. Armed with new tools and dissatisfied with available choices,
consumers want to interact with firms.
With the social Web, the number of online reviews in which people
freely express their opinions on a whole variety of topics is
constantly increasing. Opinion mining refers to a new subarea of
information retrieval and computational linguistics that identifies
and extracts the opinion and sentiments that a text expresses. It
determines critics' opinions about a given product, book review, etc.,
which are expressed on online forums, blogs, or comments. Since
opinions are very important when someone wants to consider other
viewpoints before making a decision, opinion mining has recently been
applied to a wide variety of applications in politics, government, and
marketing.
The aim of this special issue is to explore the recent advances in the
application of opinion mining technologies in the industry by asking
for original scientific contributions in the form of theoretical
foundations, case studies, techniques, tools, and applications of
sentiment analysis technologies.
The call for papers of this Special Issue was published on major
international email lists, on the home page of this journal, as well
as on the official Web pages of several universities. Editors received
a large amount of submissions that were later peer-reviewed by top
experts in the field. Based on the reviews and our reading of the
papers, 6 high-quality articles were selected for their
publication. Contributions of these papers are summarized as follows:
In the first paper, entitled "Opinion Retrieval for Twitter Using
Extrinsic Information", Yoon-Sung Kim, Young-In Song, and Hae-Chang
Rim describe a method to identify tweets subjectivity by using
extrinsic information about how a tweet is presented. The method was
validated and results show that all of the proposed features are
useful in the opinion retrieval system.
The second contribution entitled "An Aspect-Based Sentiment Analysis
Approach to Evaluating Arabic News Affect on Readers", by Mohammad
AL-Smadi, Mahmoud Al-Ayyoub, Huda Al-Sarhan, and Yaser Jararweh,
presents the use of an aspect-based sentiment analysis to evaluate how
Arabic news affect readers. The method is divided into two tasks:
aspect term extraction and aspect term polarity. This method was
validated using four different classifiers: CRF, J48, Naïve Bayes,
and IBk.
In the third paper, entitled "Feature-Based Sentiment Analysis for
Service Reviews", Ariyur M. Abirami and Abdulkhader Askarunisa propose
a sentiment classifier model using the improved Term Frequency Inverse
Document Frequency (TF-IDF) method and a linear regression model to
classify online reviews, tweets, or customer feedback for various
features. The sentiment analysis on tweets/reviews is done for various
features using Natural Language Processing (NLP) and Information
Retrieval (IR) techniques. The statistical results show that improved
TF-IDF is more accurate, if compared with TF and TF-IDF methods used
for representing the text.
The fourth paper entitled "Web Service SWePT: A Hybrid Opinion Mining
Approach", by Yolanda Raquel Baca-Gomez, Alicia Martinez, Paolo Rosso,
Hugo Estrada and Delia Irazu Hernandez Farias, presents a Web service
for polarity detection in Mexican Spanish. The service is based on a
hybrid approach that combines the Sequential Minimal Optimization
(SMO) machine learning algorithm with the use of features obtained by
an affective lexicon in Mexican Spanish and a corpus. Results show
that the method can be considerably improved by using a
specific-domain corpus instead of a general corpus.
In the fifth contribution entitled "Sentiment Classification of
Spanish Reviews: An Approach Based on Feature Selection and Machine
Learning Methods", María del Pilar Salas-Zárate, Mario André s
Paredes-Valverde, Jorge Limon-Romero, Diego Tlapa and Yolanda
Baez-Lopez introduce a method for sentiment classification of Spanish
reviews. The approach uses a hybrid feature extraction method based on
POS pattern and dependency parsing. Then, the features obtained are
semantically enriched through common-sense knowledge bases. Finally, a
feature selection method is applied to remove the noisy and irrelevant
features. The set of experiments performed in this work involved the
use of two datasets of movie reviews and technological products
domains.
The last paper entitled "Applying Brand Equity Theory to Understand
The Opinion of Consumers in Social Media", by Evangelos Kalampokisa,
Areti Karamanoua, Efthimios Tambourisa, and Konstantinos Tarabanisa,
describes an algorithm that aims at categorizing opinions based on
three marketing metrics: brand satisfaction, brand image, and purchase
intention. In particular, authors used a supervised machine learning
text classifier for each of the decision nodes of the algorithm. For
this purpose, authors collected a corpus of tweets which was manually
annotated with marketing metrics and social media metrics such as
volume and sentiment.
Finally, we, as editors, would like to express our gratitude to the
reviewers who kindly contributed to the evaluation of papers at all
stages of the editing process. We equally and especially thank
Professor Christian Gütl (Managing Editor) and Ms. Dana Kaiser
(Assistant Editor) of the Journal of Universal Computer Science
(J.UCS) for their invaluable help and for providing the opportunity to
edit this special issue.
Acknowledgements
This Special Issue is partially supported by the Research Support
Program of Fundación Séneca, Agencia de Ciencia y Tecnología
de la Región de Murcia through grants (15295/PI/10) and
(19371/PI/14). The issue has been also possible thanks to research
stays (19592/EE/14) and (19921/IV/15) funded by Fundación
Séneca, Agencia de Ciencia y Tecnología de la Región de
Murcia under the 'Jiménez de la Espada - Programa de Movilidad,
Cooperación e Internacionalización' program, within the II
PCTIRM 2011-2014 framework. Additionally, this special collection is
supported by Tecnológico Nacional de México (TECNM), the
National Council of Science and Technology (CONACYT), the Secretariat
of Public Education (SEP) through PRODEP.
Rafael Valencia-García
Ricardo Colomo-Palacios
Giner Alor-Hernández
Guest Editors
|