An Improved sentiment classification for objective word.

IJSRD - International Journal for Scientific Research & Development| Vol. 3, Issue 10, 2015 | ISSN (online): 2321-0613
All rights reserved by www.ijsrd.com 605
An Improved Sentiment Classification for Objective Word
Sangita N. Patel1
Ms. Jayna Shah2
1
M.E. Student 2
Assistant Professor
1,2
Department of Computer Engineering
1,2
Sardar Vallabhbhai Institute of Technology, Vasad-388306, Gujarat-India
Abstract— Sentiment classification is an ongoing field and
interesting area of research because of its application in
various fields. Customer sentiments play a very important
role in daily life. Currently, Sentiment classification
focused on subjective statements and ignores objective
statements which also carry sentiment. During the sentiment
classification, problem is faced due to the ambiguous sense
(meaning) of words and negation words. In word sense
disambiguation method semantic scores calculated from
SentiWordNet of WordNet glosses terms. The correct sense
of the word is extracted and determined similarity in
WordNet glosses terms. SentiWordNet extract first sense of
word which used in general sense. This work aims at
improving the sentiment classification by modifying the
sentiment values returned by SentiWordNet and compare
classification accuracy of support vector machine and naïve
bays.
Key words: Sentiment classification, Word sense
disambiguation, SentiWordNet, WordNet
I. INTRODUCTION
Text documents from various sources within the publication
of this research for the discovery of the overall opinion
shows an initial proposal. Examples of these sources, news
articles, blog posts, social media, movie review, game
reviews and product reviews. Furthermore, opinion mining
also called sentiment analysis , recent interest has gained
more than a decade, which is a relatively new field of study.
This kind of positive and negative meaning to categorize
groups wrote documents. Can also be neutral to the
mentioned but as having no opinion or cancel each other
that the same amount of positive and negative thoughts that
can be see there. Computers are getting faster on the
Internet, an increasing amount of data is becoming more
available to everyone. Very large amounts of information
are available in on-line documents.
Sentiment analysis is a process for tracking the
mood of the public about a certain product, for example, by
building a system to examine the conversations happening
around it. As part of the effort to better organize this
information for users, researchers have been actively
investigating the problem of automatic text categorization
[1]. A demand to extract and available to obtain useful
information from the data for data mining that enables
individuals and companies have risen. It is also the case with
sentiment analysis. People's opinion about companies,
movies, games, education, products and people is a way to
identify the needs. For example, companies began to brand
its new product was received in the media how the image. It
also search for relevant products to the order of the options
available to them by public opinion to give a holistic
approach can be used to help people. Gardening is talking
about, when the plants needed as human beings do not enjoy
similar weather conditions, as the sentence has meaning can
be completely opposite. Opinions can be expressed in
different ways. The following are examples of statements of
opinion.
– I read this book.
– The book is good.
– I like to read this book.
– The book is very good.
– The camera's quality is very good.
In sentiment analysis, researchers have focus
mainly on two problems in detecting whether subjective or
objective text, and subjective text is further classify as
positive or negative. Ask sense orientation, calculation and
supervised and unsupervised classification based on
machine learning techniques: Techniques used two main
approaches. Opinions varied sources, such as can be
gathered from personal conversations, newspapers,
television, Internet, etc., the Web has become the richest
source of feedback collection. Before World Wide Web
(WWW), people collected opinion manually.
Sentiment analysis has become a new knowledge
resource after the advent of the Internet and World Wide
Web. It aims automatically predict the sentiment polarity of
user's opinions on the web. Opinions play an important role
in understanding the collective sentiments and help to make
better decisions. Opinions may be positive, negative or
neutral. Positive opinions encourage the prospective
customer to take positive decision; negative opinion usually
results in negative decision. Sentiment analysis of textual
communication extracts the subjective information in the
text. The main task in sentiment classification is to
determine the polarity of the comments as positive, negative
or objective. It can be done at different levels such as
word/phrase levels, sentence level and document level.
Sentiment analysis is one of the most challenging areas in
NLP because people express opinion in subtle and complex
ways, involving the use of slang, ambiguity, sarcasm, irony
and idiom. Word Sense Disambiguation process is a process
to define the sense/meaning of an ambiguity word. The
sense of a word in a text depends on the context in which it
is used. The context is determined by the other words in the
neighborhood in the sentence. To give a hint how all this
works, consider two examples of the distinct senses that
exist for the (written) word "bass":
– a type of fish
– tones of low frequency and the sentences:
– I went fishing for some sea bass.
– The bass line of the song is too weak.
To a human, it is obvious that the first sentence is
using the word "bass (fish)", as in the former sense above
and in the second sentence, the word "bass (instrument)" is
being used as in the latter sense below[2].

(IJSRD/Vol. 3/Issue 10/2015/127)
II. LITERATURE SURVEY
This section describes literature review or the studies which
give an idea that for our research done in direction of
sentiment classification.
Yan Dang, Yulei Zhang proposed lexicon enhanced
method for sentiment classification combines machine
learning and semantic-orientation approaches into one
framework that significantly improves sentiment
classification performance. We also found that conducting
feature selection can further improve the performance,
especially for large data sets. They compared Naïve Bayes,
Maximum Entropy, and SVM and achieved the highest
classification accuracy (82.9 percent) using SVM[4].The
semantic-orientation approach, on the other hand, performs
classification based on positive and negative sentiment
words and phrases contained in each evaluation text and
mining the data requires no prior training.[4]
Chihli Hung,Hao-Kai Lin proposed approach for
mine sentiments of opinions from word-of-mouth (WOM) to
improve the performance of word-of-mouth Sentiment
classification by re-evaluates objective sentiment words in
the SentiWordNet sentiment lexicon with the help of SVM
classifier.[5] WordNet is a public sentiment lexicon that‘s
used to extract sentiments of WOM for sentiment
classification. However,most existing sentiment mining
models ignore objective words, which comprise more than
90 percent of the words in SentiWordNet. These objective
words are often considered useless. Research reevaluates
objective words in SentiWordNet by assessing the
sentimental relevance of objective words and their
associated sentiment sentences. In this paper two sampling
strategies and integrate them with the support vector
machines (SVMs) for sentiment classification.[5]
As an example, we‘ll use two sentences wherein
each word contains three sentiment values in brackets—that
is, Positive, objective, and negative—while looking up
SentiWordNet as follows:
– Sentence 1: I (p:0, o:1, n:0) will (p:0, o:1, n:0) read
(p:0, o:1,n:0) this (n/a) book (p:0, o:1, n:0) later (p:0,
o:1, n:0).
– Sentence 2: Reading (p:0, o:1, n:0) this (n/a) book (p:0,
o:1,n:0) is (n/a) happy (p:0.875, o:0.125, n:0).
A word whose sentiment value is the greatest in
positive, negative, or objective orientation is defined as a
positive, negative, or objective word, respectively.
Jasmine Bhaskar, Sruthi K.,Prema Nedungadi
proposed an enhanced technique for sentiment classification
of online reviews by considering the objective words [5] and
intensifiers[6].Intensifier Handling: People usually use
intensifiers in reviews to express their emotion deeply.
Presence of the words like 'very " 'really 'and 'extremely ' in
negative and positive sentences make the adjective and
adverb stronger. But this effect is not considered during the
score calculation in existing method. The polarity of the
sentence can be obtained by following equation.
Sentence Score= ΣI=1 Score(i)
Score(i) is the positive and negative score of the words and
n is the number of words in the sentence. If Sentence Score
is greater than 0, then we can say that the sentence is
positive otherwise sentence is negative.
M. Govindarajan proposed new hybrid
classification method is proposed based on coupling
classification methods using arcing classifier and their
performances are analyzed in terms of accuracy.[7] A
Classifier ensemble was designed using Naïve Bayes (NB),
Support Vector Machine (SVM). In the proposed work, a
comparative study of the effectiveness of ensemble
technique is made for sentiment classification. The
ensemble framework is applied to sentiment classification
tasks, with the aim of efficiently integrating different feature
sets and classification algorithms to synthesize a more
accurate classification procedure.[7]
Muhammad Faheem Khan, Aurangzeb Khan and
Khairullah Khan proposed a new method of word sense
disambiguation (WSD) using matrix map of the semantic
scores extracted from SentiWordNet of WordNet glosses
terms.[8] The correct sense of the target word is extracted
and determined for which the similarity between WordNet
gloss and context matrix is greatest. Experiment results have
shown that the proposed method improves the result of
sentence level sentiment classification as evaluated on
different domain datasets. From the result it is clear that the
propose method achieves an accuracy of 90.71% at sentence
level sentiment classification of online reviews.[8].
III. SENTIWORDNET AND WORDNET
SentiWordNet is sentiment analysis lexical resource made
up of synset from WordNet, a thesaurus-like resource; they
are allocated a sentiment score of positive, negative or
objective. These scores are automatically generated using
the semisupervised method which is described in [6]. It is
also available freely for research purpose on web.
SentiWordNet is one of the sources of sentiment analyses. It
is a semiautomatic way of providing word/term level
information on sentiment polarity by utilizing WordNet
database of English terms and relations. WordNet is is a
very rich source of lexical knowledge Since most entries
have multiple senses. Each term in WordNet database is
assigned a score of 0 to 1 in SentiWordNet which indicates
its polarity. Strong partiality information terms are assigned
with higher scores whereas less bias/subjective terms carry
low scores. SentiWordNet is made up of a semi-supervised
method which refers to a subset of seed terms to obtain
semantic polarity. Each set of synonymous terms is assigned
with three numerical scores ranging from 0 to 1 which
indicates its objectiveness i.e. positive and negative bias
[10]. One of the key features of SentiWordNet is that it
assigns both positive and negative scores for a given term
according to the following rule [6]:
Fig. 1: Graphical Representation for SentiWordNet
– Pos(s) Positive score for synsets.
– Neg(s) Negative score for synsets.

– Obj(s) Objectiveness scores for synsets.
Then the following scoring rule applies:
Pos(s) + Neg(s) + Obj(s) = 1
The positive and negative scores are always given, and
objectiveness can be implied by the relation:
Obj(s) = 1 – (Pos(s) + Neg(s))
Polarity scores according to synset and relevant part of
speech are grouped by SentiWordNet database as a text file.
IV. PROPOSED METHODOLOGY
In this work, we propose an enhanced technique for
sentiment classification of online reviews by considering the
objective words. The proposed work consists of three
modules document pre-processing, modifications of
objective words in SentiWordNet and sentiment
classification.
A. Data Pre-Processing Module Consists Following Steps
Preprocessing follows the same step as traditional text
mining which consists of sentence splitter, POS tagging and
stop word removal. Review document consists of many
sentences and each sentence expresses specific sentiment.
So sentence is considered as a basic unit here.
Fig. 2: Graphical representation for proposed method
1) Split Document into Sentences
Review document consists of several sentences and each
sentence expresses the specific emotion. Reviewing the
documents such as a comma, question mark, exclamation
point or a period, depending on the punctuation as many
sentences are first divided.
2) Part Of Speech (POS) Tagging
We use pos tagger to assign tag to each word. SentiWordNet
provides four POS : Adjective, Adverb, Verb and Noun.
Same word with different part of speech tag might have
different sentiment value. For example Word 'good' appears
in three different parts in a sentence may have different
values according to its part of speech tag. So proper part of
speech tag should be applied on each word in the sentences.
3) Stop Word Removal
Stop words are the word that doesn't carry much meaning
such as determiners and prepositions. Removal of stop
words is the last step in pre-processing.
B. Modification Of Objective Word In Sentiwordnet
Module Consists Following Steps:
In this module calculate the relevance of an objective word
and its associated sentences. The basic concept is that a
positive or a negative sentence has some sentimental
influence on its associated objective words. A positive
sentence contains greater positive value and usually has
more positive words than negative words. Thus, a positive
sentiment tag is assigned to an objective word when this
word appears in a positive sentence more often than in a
negative sentence, and vice versa.
1) Sentiwordnet For Positive, Negative Score And
Objective Score
Each word in the list updated with Positive and Negative
score from SentiWordNet lexical resource. One of the key
features of SentiWordNet is that it assigns both positive and
negative scores for a given term according to the following
rule [5,6]: For a synset s, we define.
– Pos(s) Positive score for synsets.
– Neg(s) Negative scores for synsets.
– Obj(s) Objectiveness scores for synsets.
Then the following scoring rule applies:
Pos(s) + Neg(s) + Obj(s) = 1;[5,6]
The positive and negative scores are always given, and
objectiveness Can be implied by the relation:
Obj(s) = 1 – (Pos(s) +Neg(s))[5,6]
2) Algorithm for reassigning new value to objective words
[5]
If (Objective word occurs only in positive sentence)then
PosWi=Psi/fri; NegWi=0; ObjWi=1-PosWi;
Else if(Objective word occurs only in
negative sentence)then
NegWi =Nsi/fri; PosWi =0;
ObjWi=1- NegWi;
Else if (Occurrence of Positive sentence< Occurrence of
Negative sentence) then
If (Nsi-Psi>threshold) then
NegWi =Nsi/fri; PosWi =0; ObjWi=1- NegWi;
End if
Else if (Occurrence of Positive sentence>Occurrence
of Negative sentence) then
If (Psi-Nsi>threshold)then
PosWi=Psi/fri;NegWi=0; ObjWi=1-PosWi;
End if
Else
PosWi=0;NegWi=0;ObjWi=1;
End if
3) Wordnet For Word Sense Disambiguation
The resultant score of the sentence are still not perfect
because it has ambiguity as the word of natural language can
be used in different senses having different scores according
to the context. It is very difficult to decide automatically that
which sense is used in the sentence. In this work new
method is developed for correct sense extraction during
sentiment analysis. Firstly senses of individual words are
extracts from WordNet Glosses. To select the correct sense
we create a matrix of similarity scores of word and senses
of words.Finally if correct sense is extracted then will get
the correct score of words from where the sentence score
will be improved.To extract the true sense of the word in
many senses of the word we present you with the number
and the real sense of the word are the positive and negative
scores use the WordNet glosses.
Algorithm for Word sense disambiguation [8]
1) INPUT:
Word_List:= All Sentiment Words in a sentence

AMB_Word:= Ambigues Word in a Word_List
WRD_GLOSSES_LIST: WordNet Glosses for
AMB_Word
WRD_GlOSS: AMB_Word In WRD_GLOSSES_LIST
2) OUTPUT:
W_SENSE:= WordNet Sense for Word in Word_List
ForeachAMB_Word in Word_List
SELECT W_POSITIVE_SCORE AND
W_NEGATIVE_SCORE FROM SentiWordNet
Foreach WRD_GLOSS in
WRD_GLOSSES_LIST
SELECT G_POSITIVE_SCORE AND
G_NEGATIVE_SCORE FROM
SentiWordNet
If W_POSITIVE_SCORE AND
W_NEGATIVE_SCORE is similar
G_POSITIVE_SCORE AND
G_NEGATIVE_SCORE
WRD_GLOSS then W_SESNSE:=WRD_GLOSS
End if
End for
End for
C. Sentiment Classification
1) Vector Representation Of The Document
SVM has high accuracy as compared to other machine
learning approaches At first, represent the document as a
vector Di = [WI, W2, W3 ... Wn] Where Wi is the weight of
the term i with respect to the document.
2) Sentiment Classification Using SVM
Support vector machines are a set of related supervised
learning methods that analyze data and recognize patterns,
used for classification and regression analysis. The standard
SVM is a non-probabilistic binary classifier or binary linear
classifier, i.e. it predicts, for each given input, which of two
possible classes the input is a member of. Since an SVM is a
classifier, then given a set of training examples, each
marked as belonging to one of two categories, an SVM
training algorithm builds a model that predicts whether a
new example falls into one category or the other[2]. More
formally, a support vector machine constructs a hyperplane
or set of hyperplanes in a high or infinite dimensional space,
which can be used for classification, regression or other
tasks. Intuitively, a good separation is achieved by the
hyperplane that has the largest distance to the nearest
training data points of any class, since in general the larger
the margin the lower the generalization error of the
classifier[2].
V. RESULTS
In this section, we demonstrate the performance of our
proposed method by comparing it with the existing method.
For our experiments we have used the data set taken from
Amazon.com for the products reviews of digital camera
from which we have taken 314 sentences and we have used
the data set taken from imdb.com for the moive reviews
which we have taken 121 sentences for training and testing.
The performance of the classifier can be measured in terms
of the four possible outcomes: True positive (TP), true
negative (TN), false positive (FP), and false negative
(FN).True positivenegative means that a sentence is
classified to a positivenegative class when this sentence
really belongs to the positivenegative class respectively.
Both true positive and true negative are correct
classifications. False positivenegative means that a sentence
is incorrectly classified to a negativepositive class when
this sentence belongs to a positivenegative class. Accuracy
of existing and proposed method is calculated by using the
equation given below.
Accuracy = TP+TN/TP + FP + TN + FN
From the result we can see that the prediction
accuracy of the proposed method compare with SVM and
NB in which SVM is much better than the NB. Reassigning
objective word as positive or negative, improved the
classification accuracy by reducing both positive and
negative misclassification.. This is because miss-
classification is less in the proposed method as compared to
the existing method.
Dataset Classifier Accuracy
Movie Review Dataset
SVM
77.69
NB 72.72
Product Review Dataset
SVM
81.97
NB 74.32
Table 1:
VI. CONCLUSION
We have seen that sentiment analysis has many applications
and it is important ﬁeld to study. Sentiment analysis has
Strong commercial interest because Companies want to
know how their products are being perceived and also
Prospective consumers want to know what existing users
think. In this work proposed method used to improve the
sentiment classification of product reviews by considering
the objective words and handle word sense disambiguation
problem combine. In this work we have used support vector
machine and naïve bays for sentiment classification. From
the result we can see that the prediction accuracy of the
proposed method compare SVM and NB in which SVM is
much better than the NB.
ACKNOWLEDGMENT
We take the immense pleasure in expressing our humble
note of gratitude to our project guide, Ms Jayna Shah,
Assistant Professor, Department of Computer
Engineer,Sardar vallabhbhai Institute of Technology, for
this remarkable guidance and suggestions, which helped us
in completion of paper.
REFERENCES
[1] Pang, Bo, Lillian Lee, and Shivakumar Vaithyanathan.
"Thumbs up?: sentiment classification using machine
learning techniques." Proceedings of the ACL-02
conference on Empirical methods in natural language
processing-Volume 10.Association for Computational
Linguistics, 2002.
[2] en.wikipedia.org/wiki/
[3] Turney, Peter D. "Thumbs up or thumbs down?:
semantic orientation applied to unsupervised
classification of reviews." Proceedings of the 40th
annual meeting on association for computational

linguistics. Association for Computational Linguistics,
2002.
[4] Yan Dang; Yulei Zhang; Hsin chun Chen, "A Lexicon-
Enhanced Method for Sentiment Classification: An
Experiment on Online Product Reviews, ―Intelligent
Systems, IEEE , vol.25, no.4, pp.46,53, July-Aug.2010
doi: 10.1109/MIS.2009.105
[5] Hung, Chihli, and Hao-Kai Lin. "Using objective
words in SentiWordNet to improve word-of-mouth
sentiment classification." IEEE Intelligent Systems
28.2 (2013): 0047-54.
[6] Bhaskar, J.; Sruthi, K.; Nedungadi, P., "Enhanced
sentiment analysis of informal textual communication
in social media by considering objective words and
intensifiers," Recent Advances and Innovations in
Engineering (ICRAIE), 2014 , vol., no., pp.1,6, 9-11
May 2014 doi: 10.1109/ICRAIE.2014.6909220
[7] Govindrajan M. "Sentiment classification of movie
review using hybrid method."
[8] Muhammad faheem Khan, Aurangzeb and khairullah
khan―efficient word sense disambigution teqnique for
sentence level sentiment classification of online
review‖ Sci.Int(Lahore).25(4),2013
[9] Lee, Wei Jan, and Edwin Mit. "Word sense
disambiguation by using domain knowledge."
Semantic Technology and Information Retrieval
(STAIR), 2011

An Improved sentiment classification for objective word.

Recommended

More Related Content

What's hot (20)

Viewers also liked (11)

Similar to An Improved sentiment classification for objective word. (20)

More from IJSRD (20)

Recently uploaded (20)

An Improved sentiment classification for objective word.