Opinion Mining Techniques for Non-English Languages: An Overview

Mhaske N.T. & Patil A.S.
International Journal of Computational Linguistics (IJCL), Volume (7) : Issue (2) : 2016 54
Opinion Mining Techniques for Non-English Languages: An
Overview
Mhaske N.T. neelimamhaske@gmail.com
School of Computer Sciences,
North Maharashtra University, Jalgaon,
Maharashtra, 425001, India
Patil A.S. ajaypatil.nmu@gmail.com
School of Computer Sciences,
North Maharashtra University, Jalgaon,
Maharashtra, 425001, India
Abstract
The amount of user-generated data on web is increasing day by day giving rise to necessity of
automatic tools to analyze huge data and extract useful information from it. Opinion Mining is an
emerging area of research concerning with extracting and analyzing opinions expressed in texts.
It is a language and domain dependent task having number of applications like recommender
systems, review analysis, marketing systems, etc. Early research in the field of opinion mining
has concentrated on English language. Many opinion mining tools and linguistic resources have
been built for English language. Availability of information in regional languages has motivated
researchers to develop tools and resources for non-English languages. In this paper we present a
survey on the opinion mining research for non-English languages.
Keywords: Opinion Mining, Sentiment Analysis, Natural Language Processing.
1. INTRODUCTION
The rapid growth of information over web has given rise to the need of tools and techniques to
automatically analyze this huge information and represent it in the form that will be convenient for
average users to work with. The information over web is available in different forms like numeric
data, textual data, images, multimedia contents, etc. How to process and use this information
depends on the application for which it is being used.
In this paper we discuss one of the popular areas of research - Opinion Mining / Sentiment
Analysis. Being social by nature, people value opinions of other people. What other people think
has always been a matter of concern, particularly when making decisions. Before the web, people
used to consult with friends, colleagues, family, etc. However nowadays people prefer to use
Internet for getting as well as sharing information. Many e-commerce sites provide platform for
users to express their opinions about the services and products. Newspaper sites give platform
for users to comment on news articles. The information collected from these various user
generated data provides insight into what the users think on that matter. This information is useful
for commercial industries to understand the user expectations about their services and products,
for political parties to analyze common man's opinions on their party or certain candidate, for
individuals to make decisions, etc. While using the information on web, the major problem is size
of the available data. For a single issue there might be hundreds to thousands of comments /
reviews available. Reading all this data manually and analyzing opinions is not feasible and time
efficient. Another issue is that of language. The language used in user generated contents is not
necessarily grammatically correct. Along with opinions there might also be other information like
facts, observations, etc. that needs to be separated from the opinions. The problems in manually
analyzing opinion information has lead researchers to develop automatic tools and techniques

that can efficiently process this data and produce opinion summaries. In this paper we focus on
the opinion mining work for non-English languages. The papers are chosen so as to explore
research carried out to develop resources for languages other than English.
Rest of the paper is organized as follows: Section 2 provide a brief information of the problem
definition of opinion mining. Section 3 discuss some previous work. Section 4 focus on
construction of linguistic resources used for opinion mining. Section 5 and 6 discuss opinion
mining related works in non-English and Indian languages.
2. THE PROBLEM
Opinion mining, also known as sentiment analysis is the area of research that deals with
automatic extraction and analysis of subjective information in the text. The subjective information
may be in the form of opinions, sentiments, beliefs, emotions, attitudes, etc. Opinion mining is a
language and domain dependent task. The sentiment analysis results are influenced by the
differences in grammar and usage of language [1]. This problem gives rise to the need of
developing new systems and resources for new languages. English is the most studied language
in the field of opinion mining. Many linguistic resources are available for analyzing opinions in
English language. However the increasing need of automatic opinion mining systems have
motivated many researchers to study different languages.
The document on which opinion mining is to be applied, needs to be subjective in nature. The
types of documents used for automatic opinion mining includes:
Domain Used in
Product Reviews [2][3][4][5][6][7]
Movie Reviews [8][9][10][11]
News articles [12][13][14][15][16][17]
Social Media Contents [18][19]
TABLE 1: Domains for Opinion Mining.
The workflow of opinion mining follows the steps:
 Feature Extraction - This step is applied for aspect-based opinion mining where the
opinions are extracted around a particular feature / aspect of entity under consideration.
 Subjectivity Classification - This task is related to identifying opinions and facts. The
opinion mining systems require input documents to be opinionated. However one cannot
be sure if a document contains only opinions. So input documents are first processed to
filter opinion information.
 Sentiment Analysis - Sentiment analysis is performed on subjective data. It deals with
identifying the overall polarity of subjective text.
For each of the above tasks, different methodologies have been adopted by researchers.
3. RELATED WORK
Opinion mining is a very popular and challenging research area. Many researchers are working
on opinion mining systems for different languages with different issues and challenges. The
research work in the field of opinion mining have been presented in many surveys. Pang and Lee
(2008) presented a comprehensive survey on opinion mining and sentiment analysis research
covering the application areas, issues and challenges, techniques used for performing various
intermediate steps, etc. [20]. Montoyo et al. (2012) presented an overview of current state of the
research in the field of sentiment analysis [21]. The review focus on four major categories of
approaches namely, resource construction, text classification, opinion extraction and sentiment
analysis. The detailed survey of Tsysarau and Palpanas (2012) presented an overview of web
mining and subjectivity analysis algorithms. The study reviewed the most prominent approaches

for the problems of opinion mining and opinion aggregation, contradiction analysis, etc. [22]. In
survey on opinion mining framework, Selvam and Abirami (2013) discussed the various steps
involved in the opinion mining and provided a brief listing of different techniques used in each
step [23]. The survey of Medhat et. al. (2014) provided sophisticated categorization of a large
number of techniques used for sentiment analysis [24]. They also discussed available benchmark
data sets and categorized them according to their use in certain applications. The latest review of
Rana and Cheah (2016) presented a comprehensive overview of different aspect extraction
techniques and approaches [25]. The review focus on various techniques for implicit and explicit
aspect extraction along with techniques for aspect categorization. The research in the field of
opinion mining is continuously increasing with new approaches and new languages.
Indian languages are also gaining interest among researchers to develop natural language
processing resources and systems like named entity recognition [26][27], stemmers[28], etc. The
purpose of present study is to explore the research in opinion mining that has been carried out for
the non-English languages.
4. SENTIMENT LEXICON
Development of opinion mining systems require the knowledge of how opinions are expressed in
the target language. Opinions may be expressed at word, sentence, paragraph or document
level.
The most common way to express opinions in any language is by using sentiment bearing words
like 'excellent', 'good', 'bad', etc. These words can easily be classified as positive or negative. The
collection of such sentiment bearing words and phrases provide a valuable information for
building opinion mining systems. Sentiment Lexicon is a basic lexical resource used in many
systems. The sentiment lexicon is a collection of sentiment words along with their polarity
information. The lexicons include:
1. Sentiment Word / Phrase
2. Polarity (Positive/Negative/Neutral)
3. Strength of Polarity (numeric values / degree of intensity in form of strong - weak range)
Sentiment lexicons play an important role in building classifiers that rely on the presence of
lexicon entries in the text. The sentiment lexicon may be domain dependent or domain
independent. As word senses may change according to contexts, domain oriented lexicons
produce better results for the selected domain. Manually constructing the sentiment lexicon for
new language is very time consuming and laborious task. So semi-automatic and fully automatic
techniques are used for the task. Table 2 summarizes lexicon constructions techniques used for
different languages:
Authors (Year) Language Resources
Used
Technique Polarity
Hiroshi, Tetsuya
and Hideo
(2004) [29]
Japanese Product reviews +
Bilingual lexicon
Corpus Based Extraction of
Sentiment Units
Mihalcea,
Banea
and Wiebe
(2007) [30]
Romanian SemCor corpus +
Romanian
documents
Corpus Based Projection
Mihalcea,
Banea
and Wiebe
(2007) [30]
Romanian OpinionFinder Machine
Translation
Inherited from
the English
Resource
Banea, Mihal-
cea and Wiebe
Romanian Online Romanian
dictionary
Wordnet
Based
Bootstraping

(2008) [31]
Bestgen (2008)
[32]
French French
lexicon +
Newspaper
Articles
Corpus
Based
SO-LSA
DI-LSA
Kim, et al.
(2009) [33]
Korean OpinionFinder
+ SentiWordNet
Machine
Translation
Inherited from
the English
Resource
Jijkoun and
Hofmann
(2009) [34]
Dutch Dutch
Wordnet
Wordnet
Based
PageRank
like algorithm
Waltinger U.
(2010) [35]
German Subjectivity
Clues +
SentiSpin
Machine
Translation
Inherited from
the English
Resource
Remus,
Quasthoff and
Heyer
(2010) [36]
German General Inquirer Machine
Translation
PMI technique
Abdul-Mageed
and Diab
(2012) [37]
Arabic SentiWordNet
+ Youtube
lexicon + General
Inquirer
Machine
Translation
From scores
of words in
English lexicon
Paulo-Santos,
Ramos and
Marques
(2011) [38]
Portuguese Portuguese
Wordnet
+ Common online
Dictionary
Wordnet
Based
Graph algorithm
TABLE 2: Sentiment Lexicon Construction Methods.
4.1. Machine Translation Approach
With this approach, existing resources from source language are translated into target language.
The translation approach can easily be adopted for languages for which bilingual dictionaries /
translation tools are available. As many useful resources are available for English, it is the most
preferred source language. Although many sentiment lexicons are constructed for English
language, following are the lexicons that are used by many researchers:
 Sentiwordnet
1
: it is a publicly available lexical resource for supporting sentiment
classification and opinion mining applications. It is based on Word Net synsets. The
strength of polarity is represented in the form of numeric values.
 OpinionFinder (MPQA Subjectivity Lexicon)
2
: this system aims to identify subjective
sentences and to mark various aspects of the subjectivity in these sentences. In the
subjectivity lexicon each word is assigned its part of speech category with polarity as
positive/ negative/neutral and subjectivity strength as weak or strong.
 General Inquirer
3
: is a lexicon attaching syntactic, semantic, and pragmatic information
to part-of-speech tagged words.
The polarity information for the target language lexicon can be inherited from the source
resource.
The translation based approach is simple and easy to follow. A large size lexicon can be
constructed in less time with minimum human efforts. However, the major problem with this
1 http://sentiwordnet.isti.cnr.it/
2 http://mpqa.cs.pitt.edu/lexicons/
3 http://www.wjh.harvard.edu/~inquirer/

approach is that, sense information might be lost in translation. Also there may arise another
problem like translation of multiword expressions, translation ambiguity, handling of inflections,
etc. that needs to be addressed differently for different languages [30].
4.2. Wordnet/Thesaurus Based Approach
This is another popular approach applicable to the languages for which wordnet / thesaurus is
available. Starting with an initial set of seed words with known polarity values, wordnet /
thesaurus is used to expand the initial seed set by using lexical relations and propagate the
polarity values to unknown words. With wordnet / thesaurus method a large sized lexicon can be
constructed. The lexicon may be used for general purpose as well as domain specific
applications. The method is computationally difficult and requires the availability of wordnet /
thesaurus.
4.3. Corpus Based Method
In the corpus based methods, the lexicon is constructed using corpus in target language and
language specific features. The corpus based method relies on association of words. e.g. Turney
and Littman (2003) introduced a method for inferring the semantic orientation of a word from its
statistical association with a set of positive and negative paradigm words [39]. The statistical
measures used are Pointwise Mutual Information (PMI) and Latent Semantic Analysis (LSA). This
method is more suited for domain specific applications.
5. APPROACHES TO OPINION MINING
A large number of approaches have been implemented for various tasks in the opinion mining
systems. The approaches can broadly be classified into Lexicon/Rule based and Machine
Learning based.
5.1. Lexicon/Rule Based and Statistical Approaches
The lexicon based approaches use sentiment lexicon and set of rules to extract and classify
subjective text. The language specific issues, handling of negation, identification of features etc.
are modeled in the form of rules. Most of the rule based techniques use term presence and
frequency as the basis for classification. The problem with this technique is that, quality of the
underlying lexicon affects the performance of the classifier. Sentiment lexicons suffer from
inability to consider the specific context in which the words are used [14]. To handle this problem
Ding et al. (2008) proposed a holistic approach that exploit external evidences and linguistic
conventions of natural language expressions [5].
Use of statistical measures is another popular approach to classify opinions. This approach
employ corpus and statistical measures to find co-occurrence patterns to identify and extract
opinions. Measures like PMI (Pointwise Mutual Information), LSA (Latent Semantic Analysis),
Semantic Orientation, etc. are used for this task. For efficient systems, the statistical techniques
require availability of large corpus in the target language which may not be easily available. This
approach is used for very few non-English languages. Some of the lexicon based and statistical
approaches are summarized in table 3:
Authors (Year) Language Task Technique Result
Fujii and
Ishikawa (2006)
[40]
Japanese Opinion
Summarization
Rule Based Graphical Results
Wang and Araki
(2007) [2]
Japanese Opinion Mining Modified SO-
PMI
Accuracy
(Pos:78%,
Neg:72%)
Zubaryeva and
Savoy (2009)
[17]
Japanese
Chinese
English
Opinion
Detection
Statistical
Approach using
Z Score and
Logistic Model
F1 Measure
(English:0.54,
Chinese:0.82,
Japanese:0.62)

Wang and Fu
(2010) [16]
Chinese Sentiment
Classification
Sentiment
Morphemes
based
F Measure(0.40)
Bal et al. (2011)
[1]
Dutch and
English
Multilingual
Sentiment
Analysis
Lexicon Based Accuracy
(Dutch:79%,
English:71%)
TABLE 3: Lexicon Based / Statistical Approaches to Opinion Mining Tasks.
5.2. Machine Learning Based Approaches
In machine learning approaches the opinion mining task is viewed as text classification problem
and various machine learning algorithms are employed using different features. This approach
involves construction of training data for the classification task. The supervised approaches use
data annotated with sentiment labels while the unsupervised approaches use unannotated data.
The next step is to represent the training data in the form of feature vectors which are used to
train the classifier. Finally, the trained classifier is used to predict opinions in unseen documents /
texts. Table 4 summarizes machine learning techniques adopted for different languages.
Authors (Year) Language Task Technique
4
Result
Kobayashi et al.
(2005) [3]
Japanese Opinion
Extraction
SVM F-measure (Attr.-
Value Pairs:58)
Wang and Araki
(2007) [7]
Japanese Opinion
Mining
NB + modified
SO-PMI
Graphical Results
Kanamaru et al.
(2007) [13]
Japanese Opinion
Extraction
SVM Accuracy (Opin-
ion Extrac-
tion:42.88%,
Op. holder ex-
traction:14.31%,
Polarity:19.90%,
Relevance judg-
ment:63.15%)
Atteveldt et al.
(2008) [14]
Dutch Sentiment
Analysis
ME F1 Score(0.63)
Zhang et al.
(2008) [41]
Chinese Sentiment
Classification
Lexicon Based
+
SVM + NB + DT
Accuracy
(SVM:81.65%,
Lexicon
Based:76.26%)
Mellebeek et al.
(2010) [6]
Spanish Opinion
Mining
Classifiers
implemented in
Mallet and
Weka
Use of non-expert
annotations
through crowd
sourcing is a
viable and cost
effective
alternative to the
use of expert
annotations
Martinez-Camara
et al. (2011) [10]
Spanish Opinion
Classification
SVM+NB+BBR+
KNN+C4.5
Highest Precision
with
BBR (87%)
4
SVM-Support Vector Machine, NB-Naive Bayes, ME-Maximum Entropy, DT-Decision Tree, BBR-Bayesian
Binary Regression, KNN-K Nearest Neighbor

Lee and Renganathan
(2011) [4]
Chinese Sentiment
Analysis
ME Accuracy (0.87%)
Rushdi- Saleh et al.
(2011) [9]
Arabic +
English
Opinion
Mining
SVM + NB F1 Measure
(SVM without
stemmer:0.90%)
Hamouda and El-taher
(2013) [15]
Arabic Sentiment
Analyzer
DT + SVM + NB Accuracy
(SVM:73.4%)
Yang and Chao
(2014) [11]
Chinese Sentiment
Analysis
PMI + SVM Average
balanced
accuracy rate
(0.7)
Duwairi and Qarqaz
(2014) [19]
Arabic Sentiment
Analysis
SVM + NB +
KNN
Highest Precision
(SVM:75.25%),
Highest Recall
(KNN: K=10,
69.04%)
Ibrahim et al.
(2015) [18]
Arabic Sentiment
Analysis
SVM Accuracy (95%)
TABLE 4: Machine Learning Techniques for Opinion Mining.
The quality and size of training data highly affect the performance of the classifier. While working
with machine learning algorithms it is necessary to define useful features for the training and
testing instances. The selection of features affects the classification results. The most commonly
used features for opinion mining tasks are bag of words, appraisal phrases, n-grams, sentiment
words, term frequency, term presence, sentiment word position, valence shifters, etc. Although
many types of machine learning algorithms have been experimented for the opinion mining task,
the two popular algorithms used in English as well as non-English languages are:
Support Vector Machines [3][9][10][11][13][15][18][19][41]
It is a supervised binary classifier that takes as
input set of examples with associated class
labels. The machine construct a hyperplane
that separate the classes. The new examples
are categorized based on this hyperplane.
Naive Bayes [7][9][10][15][19][41]
It is a simple probabilistic classifier that
estimates the class-conditional probability by
assuming that the attributes are conditionally
independent. The algorithm make use of Bayes
Theorem to predict the probability of unknown
instances.
TABLE 5: Popular Machine Learning Algorithms for Opinion Mining.
6. OPINION MINING FOR INDIAN LANGUAGES
Although many non-English languages are being studied for building automated opinion mining
systems, very little work in this field has been reported for Indian languages. Das and
Bandyopadhyay (2009) initiated a study in sentiment analysis of Bengali language on the news
and blog corpus [12]. Using the translation approach sentiment lexicon was constructed and rule
based technique was employed for subjectivity annotation. The final classifier was built using
conditional random fields that resulted in precision values of 72.16% and 74.6% for the news and
blog domains respectively. Continuing the work, they implemented single-document opinion

summarization system for Bengali with topic based document-level theme relational graphical
representation [42].
For Hindi language, Bakliwal et al. (2012) proposed a graph based method to generate the Hindi
subjectivity lexicon [43]. The Hindi WordNet, synonym and antonym relations and simple graph
traversal were exploited to construct the subjectivity lexicon. The proposed algorithm achieved
79% accuracy on classification of reviews and 70.47% agreement with human annotators.
Another lexical resource for Hindi, Hindi-SentiWordNet (H-SWN), presented by Joshi et al. (2010)
was constructed by linking Hindi wordnet with English SentiWordNet [8]. To evaluate the lexicon,
three techniques viz. in language, machine translation based and resource-based sentiment
analysis were applied on movie reviews. Out of the three techniques, In-language approach out
performed others with accuracy of 78.14%.
Kaur and Gupta (2014) proposed a lexicon based algorithm of sentiment analysis for Punjabi text
[44]. Translation based approach was adopted for lexicon construction with Hindi Subjectivity
lexicon as the source lexicon. The resulting system achieved F1 score of 0.67.
Mhaske and Patil (2016) reported various issues and challenges in analyzing opinions in Marathi
text [45]. Other languages are also being explored for developing opinion mining resources and
systems.
7. CONCLUSION
Research in development of opinion mining systems is continuously increasing with the growing
availability of subjective data in different languages. With new approaches, new issues and
challenges get emerged. Although the English language has dominance in the field of opinion
mining, developing opinion mining resources for non-English languages is also gaining interest
among researchers.
This paper presented an overview of the opinion mining tasks and techniques implemented to
construct opinion mining resources and systems for non-English languages. The purpose of the
present study is to review different techniques employed for different languages so as to provide
direction to develop resources and systems for new languages. The area of opinion mining has
attracted many researchers due to its practical applications and need to automate the analysis
process. The major challenge in opinion mining of non-English languages is the unavailability of
linguistic resources. Present review discuss various methodologies adopted by researchers to
develop opinion mining related resources for different languages. The study will be helpful to
decide on the adoption of technique to be applied for new languages.
8. REFERENCES
[1] D. Bal, M. Bal, A. Bunningen, A. Hogenboom, F. Hogenboom, and F. Frasincar. “Sentiment
Analysis with a Multilingual pipeline," LNCS, no. 6997, pp. 129-142, 2011.
[2] G. Wang and K. Araki. “Modifying SO-PMI for Japanese Weblog Opinion Mining by using a
Balancing Factor and Detecting Neutral Expressions," in Proceedings of NAACL HLT 2007,
Companion Volume, Association for Computational Linguistics, 2007, pp. 189-192.
[3] N. Kobayashi, R. Iida, K. Inui, and Y. Matsumoto. “Opinion Extraction Using a Learning-
based Anaphora Resolution Technique," in The Second International Joint Conference on
Natural Language Processing (IJCNLP), Companion Volume to the Proceeding of
Conference including Posters/Demos and Tutorial Abstracts, 2005, pp. 175-180.
[4] H. Y. Lee and H. Renganathan. “Chinese Sentiment Analysis using Maximum Entropy," in
Proceedings of the Workshop on Sentiment Analysis where AI meets Psychology (SAAIP),
IJCNLP, 2011, pp. 89-93.

[5] X. Ding, B. Liu, and P. S. Yu. “A Holistic Lexicon-Based Approach to Opinion Mining," in
Proceedings of the 2008 International Conference on Web Search and Data Mining,
WSDM '08, ACM, 2008, pp. 231-240.
[6] B. Mellebeek, F. Benavent, J. Grivolla, J. Codina, M. R. Costa-jussa, and R. Banchs.
“Opinion Mining of Spanish Customer Comments with Non-Expert Annotations on
Mechanical Turk," in Proceedings of the NAACL HLT 2010 Workshop on Creating Speech
and Language Data with Amazon's Mechanical Turk, Association for Computational
Linguistics, 2010, pp. 114-121.
[7] G. Wang and K. Araki. “OMS-J: An Opinion Mining System for Japanese Weblog Reviews
using a Combination of Supervised and Unsupervised Approaches," in NAACL HLT
Demonstration Program, Association for Computational Linguistics, 2007, pp. 19-20.
[8] A. Joshi, A. R. Balamurali, and P. Bhattacharyya. “A Fall-Back Strategy for Sentiment
Analysis in Hindi: A Case Study," in Proceedings of ICON 2010: 8th International
Conference on Natural Language Processing, Macmillan Publishers, 2010.
[9] M. Rushdi-Saleh, M. T. Martin-Valdivia, L. A. Urena-Lopez, and J. M. Perea-Ortega.
“Bilingual Experiments with an Arabic-English Corpus for Opinion Mining.," in Proceedings
of Recent Advances in Natural Language Processing, 2011, pp. 740-745.
[10] E. Martinez-Camara, M. T. Martin-Valdivia, and L. A. Urena-Lopez. “Opinion Classification
Techniques applied to a Spanish Corpus," LNCS, vol. 6716, pp. 169-176, 2011.
[11] H. Yang and A. Chao. “Sentiment Analysis for Chinese Reviews of Movies in Multi-Genre
based on Morpheme-based Features and Collocations," Information Systems Frontiers,
vol. 17, no. 6, pp. 1335-1352, 2014.
[12] A. Das and S. Bandyopadhyay. “Subjectivity Detection in English and Bengali: A CRF-
based Approach," in Proceedings of ICON-2009: 7th International Conference on Natural
Language Processing, Macmillan Publishers, 2009.
[13] T. Kanamaru, M. Murata, and H. Isahara. “Japanese Opinion Extraction System for
Japanese Newspapers using Machine-Learning Method," in Proceedings of NTCIR-6
Workshop Meeting, 2007, pp. 301-307.
[14] W. Atteveldt, J. Kleinnijenhuis, N. Ruigrok, and S. Schlobach. “Good News or Bad News?
Conducting Sentiment Analysis on Dutch Text to Distinguish between Positive and
Negative Relations," Journal of Information Technology & Politics, vol. 5, no. 1, pp. 73-94,
2008.
[15] A. Hamouda and F. El-taher. “Sentiment Analyzer for Arabic Comments System,"
International Journal of Advanced Computer Science and Applications, vol. 4, no. 3, pp. 99-
103, 2013.
[16] X. Wang and G. Fu. “Chinese Sentence-Level Sentiment Classification based on
Sentiment Morphemes," in International Conference on Asian Language Processing, IEEE
Computer Society, 2010, pp. 203-206.
[17] O. Zubaryeva and J. Savoy. “Investigation in Statistical Language-Independent
Approaches for Opinion Detection in English, Chinese and Japanese," in Proceedings of
CLIAWS3, Third International Cross Lingual Information Access Workshop, Association for
Computational Linguistics, 2009, pp. 38-45.
[18] H. S. Ibrahim, S. M. Abdou, and M. Gheith. “Sentiment Analysis for Modern Standard
Arabic and Colloquial," International Journal on Natural Language Computing, vol. 4, no. 2,
pp. 95-109, 2015.

[19] R. M. Duwairi and I. Qarqaz. “Arabic Sentiment Analysis using Supervised Classification,"
in The 1st International Workshop on Social Networks Analysis, Management and Security,
2014, pp. 1-10.
[20] B. Pang and L. Lee. “Opinion Mining and Sentiment Analysis," Foundations and Trends in
Information Retrieval, vol. 2, no. 1-2, pp. 1-135, 2008.
[21] A. Montoyo, P. Martinez-Barco, and A. Balahur. “Subjectivity and Sentiment Analysis: An
Overview of the Current State of the Area and Envisaged Developments," Decision Support
Systems, vol. 53, no. 4, pp. 675-679, 2012.
[22] M. Tsytsarau and T. Palpanas. “Survey on Mining Subjective Data on the Web," Data
Mining and Knowledge Discovery, vol. 24, no. 3, pp. 478-514, 2012.
[23] B. Selvam and S. Abirami. “A Survey on Opinion Mining Framework," International Journal
of Advanced Research in Computer and Communication Engineering, vol. 2, no. 9, pp.
3544-3549, 2013.
[24] W. Medhat, A. Hassan, and H. Korashy. “Sentiment Analysis Algorithms and Applications:
A Survey," Ain Shams Engineering Journal, vol. 5, pp. 1093-1113, 2014.
[25] T. A. Rana and Y. Cheah. “Aspect Extraction in Sentiment Analysis: Comparative Analysis
and Survey," Artificial Intelligence Review, pp. 1-25, 2016.
[26] N. Patil, A. Patil, and B. Pawar, “Survey of Named Entity Recognition Systems with respect
to Indian and Foreign Languages”, International Journal of Computer Applications, vol. 134,
no. 16, pp.21-26, 2016.
[27] N. Patil, A. Patil, and B. Pawar, “Issues and Challenges in Marathi Named Entity
Recognition”, International Journal on Natural Language Computing (IJNLC), vol. 5, no. 1,
pp.15-30, 2016.
[28] H. Patil, B. Pawar, and A. Patil, “A Comprehensive Analysis of Stemmers Available for Indic
Languages”, International Journal on Natural Language Computing (IJNLC), vol. 5, no. 1,
pp.45-55, 2016.
[29] K. Hiroshi, N. Tetsuya, and W. Hideo. “Deeper Sentiment Analysis using Machine
Translation Technology," in Proceedings of the 20th International Conference on
Computational Linguistics, COLING '04, Association for Computational Linguistics, 2004.
[30] R. Mihalcea, C. Banea, and J. Wiebe. “Learning Multilingual Subjective Language via
Cross-Lingual Projections," in Proceedings of the 45th Annual Meeting of the Association of
Computational Linguistics, Association for Computational Linguistics, 2007, pp. 976-983.
[31] C. Banea, R. Mihalcea, and J. Wiebe. “A Bootstrapping Method for Building Subjectivity
Lexicons for Languages with Scarce Resources," in Proceedings of the Sixth International
Conference on Language Resources and Evaluation (LREC'08), European Language
Resources Association (ELRA), 2008.
[32] Y. Bestgen. “Building Affective Lexicons from Specific Corpora for Automatic Sentiment
Analysis," in Proceedings of the Sixth International Conference on Language Resources
and Evaluation, 2008, pp. 496-500.
[33] J. Kim, H. Jung, S. Nam, Y. Lee, and J. Lee. “Found in Translation: Conveying Subjectivity
of a Lexicon of one Language into another using a Bilingual Dictionary and A Link Analysis
Algorithm," ICCPOL 2009, LNAI 5459, 2009, pp. 112-121.

[34] V. Jijkoun and K. Hofmann. “Generating a non-English Subjectivity Lexicon: Relations that
Matter," in Proceedings of the 12th Conference of the European Chapter of the ACL,
Association for Computational Linguistics, 2009, pp. 398-405.
[35] U. Waltinger. “GermanPolarityClues: A Lexical Resource for German Sentiment Analysis,"
in Proceedings of the Seventh International Conference on Language Resources and
Evaluation (LREC), 2010, pp. 1638-1642.
[36] R. Remus, U. Quasthoff, and G. Heyer. “SentiWS - a Publicly available German-Language
Resource for Sentiment Analysis," in Proceedings of the Seventh International Conference
on Language Resources and Evaluation (LREC'10), European Language Resources
Association (ELRA), 2010.
[37] M. Abdul-Mageed and M. T. Diab. “Toward Building a Large-Scale Arabic Sentiment
Lexicon," in Proceedings of the Sixth International Global Wordnet Conference, 2012, pp.
18-22.
[38] A. Paulo-Santos, C. Ramos, and N. C. Marques. “Determining the Polarity of Words
through a Common Online Dictionary," EPIA 2011, LNAI 7026, 2011, pp. 649-663.
[39] P. D. Turney and M. L. Littman. “Measuring Praise and Criticism: Inference of Semantic
Orientation from Association," ACM Trans. Inf. Syst., vol. 21, no. 4, pp. 315-346, 2003.
[40] A. Fujii and T. Ishikawa. “A System for Summarizing and Visualizing Arguments in
Subjective Documents: Toward Supporting Decision Making," in Proceedings of the
Workshop on Sentiment and Subjectivity in Text, Association for Computational Linguistics,
2006, pp. 15-22.
[41] C. Zhang, D. Zeng, Q. Xu, X. Xin, W. Mao, and Fei-Yue. “Polarity Classification of Public
Health Opinions in Chinese," ISI 2008 Workshops, LNCS 5075, 2008, pp. 449-454.
[42] A. Das and S. Bandyopadhyay. “Opinion Summarization in Bengali: A Theme Network
Model," in IEEE International Conference on Social Computing/PASSAT, IEEE Computer
Society, 2010, pp. 675-682.
[43] A. Bakliwal, P. Arora, and V. Varma. “Hindi Subjective Lexicon: A Lexical Resource for
Hindi Polarity Classification," in 24th International Conference on Computational Linguistics
(CICLing), 2012, pp. 1189-1196.
[44] A. Kaur and V. Gupta. “Proposed Algorithm of Sentiment Analysis for Punjabi Text,"
Journal of Emerging Technologies in Web Intelligence, vol. 6, no. 2, pp. 180-183, 2014.
[45] N. Mhaske and A. Patil. “Issues and Challenges in Analyzing Opinions in Marathi Text,"
IJCSI International Journal of Computer Science Issues, vol. 13, no. 2, pp. 19-25, 2016.

Opinion Mining Techniques for Non-English Languages: An Overview

More Related Content

What's hot (18)

Similar to Opinion Mining Techniques for Non-English Languages: An Overview (20)

Recently uploaded (20)

Opinion Mining Techniques for Non-English Languages: An Overview