A Survey on Sentiment Analysis and Opinion Mining
1 like306 views
The document reviews various techniques and approaches for sentiment analysis and opinion mining, highlighting the importance of these methods in understanding users' opinions and emotions from social media data. It discusses machine learning and lexicon-based approaches in sentiment classification, detailing their applications in consumer markets, politics, and business intelligence. The paper also presents a comparative analysis of existing methodologies along with their advantages and drawbacks.
![IJSRD - International Journal for Scientific Research & Development| Vol. 3, Issue 10, 2015 | ISSN (online): 2321-0613
All rights reserved by www.ijsrd.com 1029
A Survey on Sentiment Analysis and Opinion Mining
Pooja C. Sangvikar
Department of Information Technology
Smt.Kashibai Navale College of Engineering Pune, India
Abstract— In Today’s world, the social media has given
web users a place for expressing and sharing their thoughts
and opinions on different topics or events. For this purpose,
the opinion mining has gained the importance. Sentiment
classification and Opinion Mining is the study of people’s
opinion, emotions, attitude towards the product, services,
etc. Sentiment Analysis and Opinion Mining are the two
interchangeable terms. There are various approaches and
techniques exist for Sentiment Analysis like Naïve Bayes,
Decision Trees, Support Vector Machines, Random Forests,
Maximum Entropy, etc. Opinion mining is a useful and
beneficial way to scientific surveys, political polls, market
research and business intelligence, etc. This paper presents a
literature review of various techniques used for opinion
mining and sentiment analysis.
Key words: Sentiment Analysis, Opinion Mining, Sentiment
Classification
I. INTRODUCTION
Data Mining is one of the important step of the “Knowledge
Discovery in Databases” processes or KDD, which
discovers patterns from large datasets. Also, it includes the
techniques and methods of artificial intelligence, machine
learning and statistics. To extract the valuable information
from a dataset and transform it into an understandable and
simple structure, is the main purpose of data mining process.
Opinion mining or sentiment analysis is to analyze
the users’ opinions or thoughts which are in the form of
unstructured data. The interpretation of meanings of views is
a terrible task, for that Sentiment Analysis plays a vital role.
To interpret and understand the person's views, emotions
and thoughts, the system must be made reliable and
efficient.
There are two techniques used in Sentiment
Classification.
Machine Learning Approach
Lexicon Based Approach
Machine Learning techniques include supervised
and unsupervised learning approaches. Supervised learning
consists of some classifiers like Decision tree classifiers,
Liner classifiers, Rule-based classifiers and Probabilistic
classifiers. SVM and Neural Networks are the linear
classifiers whereas Naïve Bayes, Bayesian Network, and
Maximum Entropy is Probabilistic classifiers. Lexicon
Based approaches are classified into Dictionary based and
Corpus-based methods. The corpus-based approach further
divided into the statistical and semantic approach.
The analysis of sentiments can be Document based
where the opinion in the entire document is summarized or
classified as positive, negative or neutral, other is Sentence
based where various sentences in the text are classified or
analyzed and Lastly, Sentiment Analysis can also be Phrase
level based where the polarity is used, and classification is
done. Various applications uses the Sentiment Analysis like
in consumer market for product reviews, social media for
finding general opinions about recent topics, to know the
consumer trends etc.
The remainder of this paper is organized as
follows: Section 2 gives an overview of related work already
done on Sentiment Analysis. Section 3 includes result and a
comparative table of existing systems. Section 4 concludes
survey.
II. RELATED WORK
Sentiment Analysis has been done using different techniques
or methods. Some works extract the meaning of the text,
document or sentence level while others obtain connections
between users to assign sentiment polarity for sentiment
analysis. In particular, sentiment analysis of tweets has been
done not only using approaches based on text, i.e. lexicon
based classifiers, but also by combining Natural Language
Processing and Machine Learning techniques.
Tang and Qin in [1] proposed A Joint
Segmentation and Classification Framework for Sentence
Level Classification. Here a candidate generation model is
developed to produce segmentation candidates of a sentence
and a segmentation ranking model used to score the
usefulness of segmentation candidates for sentiment
classification and also classification model for predicting the
sentiment polarity. The experiments are performed on two
datasets: a tweet dataset and a movie review dataset. The
effectiveness of this approach is verified by applying it to
these two datasets.
Neethu and Rajasri in [2], analyzed the Twitter
posts about electronic products like mobiles, laptops, etc.
using Machine Learning approach. A new feature vector is
developed for classifying the tweets as positive, negative
and extracts peoples’ opinion about products. Classifiers
like Nave Bayes, SVM, Maximum Entropy and Ensemble
classifiers are used for classification. All these different
classifiers have the same accuracy for the new feature
vector. This feature vector performs well for electronic
products.
Sentiment Analysis of Twitter has become one of
the most important research Topic. Lek and Danny in [3]
proposed an aspect-based sentiment classification approach
to analyzing opinions for tweets. Different experiments are
performed on the aspect-based sentiment classifier, to
improve the performance of existing classifiers. The
experimental results show that aspect level classifier
approach outperforms existing methods. From experimental
results, it suggests that a layered classification approach that
uses the aspect-based classifier as the first layer
classification and the tweet-level classifier as the second
layer classification is more efficient than a classifier trained
using target-dependent features. Also, this approach
consistently improves the performance of existing sentiment
classifiers.
Xia, Xu in [4] has proposed a model called Dual
Sentiment Analysis (DSA). By creating sentiment reversed
review for each training and test review, data expansion](https://image.slidesharecdn.com/ijsrdv3i100635-160224092035/85/A-Survey-on-Sentiment-Analysis-and-Opinion-Mining-1-320.jpg)
![A Survey on Sentiment Analysis and Opinion Mining
(IJSRD/Vol. 3/Issue 10/2015/235)
All rights reserved by www.ijsrd.com 1030
technique is proposed. Also, DSA framework is developed
for sentiment classification. Multi-domain datasets are used
for the experiment. Classification accuracy is compared with
Naïve Bayes, Linear SVM, etc.
Topics in Twitter are more diverse and always
dedicated themselves to a particular domain or subject. Liu,
Li, and Cheng in [5], proposed an adaptive multiclass SVM
model that transfers an initial common sentiment classifier
to a topic adaptive. An iterative algorithm is proposed which
is followed by three steps: optimization, unlabeled data
selection and adaptive feature expansion steps. Comparison
is done with the well-Known supervised sentiment
classifiers like SVM, Decision Tree, Random Forest etc. and
semi-supervised approaches. Accuracy of algorithm
increased averagely 74% on the 6 topics from public tweet
corpus.
There are different methods to analyze information
or text from particular document. But to extract and analyze
the valuable information from text messages from social
sites, Batool, Khattak in [6] proposed the Precise tweet
classification technique. To find keywords from messages,
feature extraction is used. After that semantic based filtering
of data is done to categorize the data into specific category.
By doing filtering on data, the information gain is increased
i.e. knowledge enhancer is used to avoid information loss.
To identify the interest of people’s or opinion
towards social events, Zhou, Yong in [7] proposed a Tweet
Sentiment Analysis Model (TSAM). Evaluation of TSAM is
done on political dataset i.e. Australian Federal Election
2010. The proposed model provides fast and less expensive
alternative approach over traditional polls for mining public
views. The drawback of the approach is POS processing is
not done. Also, Emotion Analysis is not included during
Sentiment Analysis.
To analyze the data, one can use Support Vector
Machine (SVM) which is supervised learning method, but
Basari, Hussain in [8] proposed a hybrid method of SVM
and Particle Swarm Optimization (PSO). To solve the
problem of dual optimization, hybrid method is proposed.
For extraction of more features, from prepared datasets,
PSO is used. The accuracy of SVM is affected by the
hybridization of SVM-PSO and which is 77%. The other
variation of SVM can be used for improving the accuracy.
To perform Sentiment Analysis on News articles
which gives fine grained analysis, Raina in [9] used Sentic
Computing approach. For Concept-Level analysis,
SenticNet; a publicly available resource is used. The
performance of engine is measured by evaluation parameters
like accuracy, precision, recall and F-measure. The proposed
approach gives an accuracy 71.2%. This Method is only
feasible for news articles i.e. it should be commercially
viable. The drawback found is verb + noun pair is not
included in Semantic Parser. So Semantic Parser can be
improved by extracting verb + noun pair.
III. RESULT AND COMPARISON
By above related work, comparison of different mechanism
used for the sentiment classification is done.
The Overview of various existing techniques,
methods and algorithms are described in the following table.
The Paper title, Datasets used, advantages, disadvantages
and accuracy are included in the table.
Paper title Datasets Used
Techniques/
Algorithms
Advantages Drawbacks Accuracy
1) A joint
segmentation and
classification
framework for
sentence level
sentiment
classification.
Tweet dataset
and movie
review dataset.
Training
algorithm and
prediction
algorithm.
-A joint modelling
framework
outperforms over
pipeline method in
various experimental
settings.
-In joint framework
static datasets are
used i.e. publicly
available dataset.
81%.
2) Sentiment analysis
in twitter using
machine learning
techniques
Tweets of
electronic
products.
Naïve Bayes ,
SVM ,
Max entropy.
-Simpler & efficient
by using machine
learning techniques.
-Performs well for
electronic products.
- Domain
dependent i.e. only
feasible for
electronic products.
NB-89.5%
SVM, Max
Entropy-
90%
3) Aspect based
twitter sentiment
Classification.
Stanford Twitter
Sentiment
(STS) consists
of 1.6 million
tweets.
Aspect based
sentiment
classification.
Aspect based
classifier consistently
increasing the
performance of
classifier.
-In the aspect based
classification,
Contextual
information is not
considered.
84%
4) Dual Sentiment
Analysis
considering two
sides of one review.
Chinese Hotel,
Notebook, 4
English Dataset.
Dual Sentiment
Analysis (DSA)
model.
-Effective for
polarity
classification.
- Strengthen the DSA
algorithm by data
expansion technique.
- In the Dual
Sentiment
Analysis, Complex
polarity shift
patterns are not
used.
73%
5) Adaptive Co-
training SVM for
Sentiment
Classification of
tweets.
Sanders Twitter
Sentiment
Corpus &
presidential
debate corpus.
Multiclass SVM,
Adaptive co-
training
algorithm.
Adaptive co-training
SVM expands topic
adaptive feature.
-Less features are
extracted.
-Timeline
visualization of
result is not done.
74%](https://image.slidesharecdn.com/ijsrdv3i100635-160224092035/85/A-Survey-on-Sentiment-Analysis-and-Opinion-Mining-2-320.jpg)
![A Survey on Sentiment Analysis and Opinion Mining
(IJSRD/Vol. 3/Issue 10/2015/235)
All rights reserved by www.ijsrd.com 1031
6) Precise Tweet
Classification and
Sentiment Analysis.
Diabetic
relevant
Dataset.
Precise Tweet
Classification
Model.
-This is a method to
collect more precise
information about
specific
Topic.
- 72.8%
7) Sentiment Analysis
on Tweets for
Social Events.
Australian
Federal Election
2010.
Tweet Sentiment
Analysis Model
(TSAM).
Knowledge enhancer
is used to avoid
information loss.
-POS processing is
not done.
-Also, Emotion
Analysis is not
included during
Sentiment
Analysis.
-
8) Opinion Mining of
Movie Review
using Hybrid
Method of Support
Vector Machine
and Particle Swarm
Optimization.
Movie review
dataset.
Support Vector
Machine and
Particle Swarm
Optimization
(SVM-PSO).
-Method solves the
problem of dual
optimization.
-Hybridization gives
better result.
-To increase the
accuracy, SVM
with other
optimization
methods can be
used.
77%
9) Sentiment Analysis
in News Articles
Using Sentic
Computing.
Multi-
Perspective
Question
Answering
(MPQA)
corpus, a large
corpus of news
articles.
News-focused
opinion-mining
engine,
sentic computing-
based
Approach.
-Method is reliable
For identifying
neutral sentences.
-only feasible for
news articles i.e. it
should be
commercially
viable.
-verb + noun pair is
not included in
Semantic Parser.
71.2%
Table 1: Overview of various existing techniques and algorithms
IV. CONCLUSION
Internet and Web technologies are continuously growing
and expanding, so the space and scope in the area of
information retrieval is also expanding. For that, Sentiment
Analysis and Opinion Mining is considered as an interesting
area of research due to its many applications in society.
Above survey explore different techniques, also these
Opinion Mining techniques suffers from a number of
problems, such as accuracy, scalability, quality, standard of
data, domain dependency, etc. These problems have to be
tackled separately and those solutions can be used to
improve the methods to do sentiment analysis and
classification.
REFERENCES
[1] Duyu Tang, Bing Qin, Furu Wei, Li Dong, Ting Liu,
and Ming Zhou, “A joint Segmentation and
Classification Framework For Sentence Level
Classification”, IEEE/ACM TRANSACTIONS ON
AUDIO, SPEECH, AND LANGUAGE PROCESSING,
VOL. 23, NO. 11, NOVEMBER 2015.
[2] Neethu M S, Rajasree R, “Sentiment Analysis in
Twitter using Machine Learning Techniques”, 4th
ICCCNT 2013 ,July 4 - 6, 2013, Tiruchengode, India.
[3] Hsiang Hui Lek and Danny C.C. Poo, “Aspect-based
Twitter Sentiment Classification”, 2013 IEEE 25th
International Conference on Tools with Artificial
Intelligence.
[4] Rui Xia, Feng Xu, Chengqing Zong, Qianmu Li, Yong
Qi, and Tao Li, “Dual Sentiment Analysis: Considering
Two Sides of One Review”, IEEE Transactions on
Knowledge and Data Engineering, MANUSCRIPT ID,
1041-4347 (c) 2015 IEEE.
[5] Shenghua Liu, Fuxin Li, Fangtao Li, Xueqi Cheng,
Huawei Shen, “Adaptive Co-Training SVM for
Sentiment Classification on Tweets”, Copyright 2013
ACM 978-1-4503-2263-8/13/10.
[6] Rabia Batool, Asad Masood Khattak, Jahanzeb
Maqbool and Sungyoung Lee, “Precise Tweet
Classification and Sentiment Analysis”, 978-1-4799-
0174-6/13/$31.00 ©2013 IEEE.
[7] Xujuan Zhou, Xiaohui Tao, Jianming Yong, ”
Sentiment Analysis on Tweets for Social
Events”,2013 IEEE 17th International Conference on
Computer Supported Cooperative Work in Design, 978-
1-4673-6085-2/13/$31.00 ©2013 IEEE.
[8] Abd. Samad Hasan Basaria,*, Burairah Hussina, I.
Gede Pramudya Anantaa, Junta Zeniarjab, “Opinion
Mining of Movie Review using Hybrid Method of
Support Vector Machine and Particle Swarm
Optimization”, 1877-7058 © 2013 The Authors.
Published by Elsevier Ltd.
[9] Prashant Raina, “Sentiment Analysis in News Articles
Using Sentic Computing”, 2013 IEEE 13th
International Conference on Data Mining Workshops,
© 2013 IEEE.](https://image.slidesharecdn.com/ijsrdv3i100635-160224092035/85/A-Survey-on-Sentiment-Analysis-and-Opinion-Mining-3-320.jpg)
A Survey on Sentiment Analysis and Opinion Mining
- 1. IJSRD - International Journal for Scientific Research & Development| Vol. 3, Issue 10, 2015 | ISSN (online): 2321-0613 All rights reserved by www.ijsrd.com 1029 A Survey on Sentiment Analysis and Opinion Mining Pooja C. Sangvikar Department of Information Technology Smt.Kashibai Navale College of Engineering Pune, India Abstract— In Today’s world, the social media has given web users a place for expressing and sharing their thoughts and opinions on different topics or events. For this purpose, the opinion mining has gained the importance. Sentiment classification and Opinion Mining is the study of people’s opinion, emotions, attitude towards the product, services, etc. Sentiment Analysis and Opinion Mining are the two interchangeable terms. There are various approaches and techniques exist for Sentiment Analysis like Naïve Bayes, Decision Trees, Support Vector Machines, Random Forests, Maximum Entropy, etc. Opinion mining is a useful and beneficial way to scientific surveys, political polls, market research and business intelligence, etc. This paper presents a literature review of various techniques used for opinion mining and sentiment analysis. Key words: Sentiment Analysis, Opinion Mining, Sentiment Classification I. INTRODUCTION Data Mining is one of the important step of the “Knowledge Discovery in Databases” processes or KDD, which discovers patterns from large datasets. Also, it includes the techniques and methods of artificial intelligence, machine learning and statistics. To extract the valuable information from a dataset and transform it into an understandable and simple structure, is the main purpose of data mining process. Opinion mining or sentiment analysis is to analyze the users’ opinions or thoughts which are in the form of unstructured data. The interpretation of meanings of views is a terrible task, for that Sentiment Analysis plays a vital role. To interpret and understand the person's views, emotions and thoughts, the system must be made reliable and efficient. There are two techniques used in Sentiment Classification. Machine Learning Approach Lexicon Based Approach Machine Learning techniques include supervised and unsupervised learning approaches. Supervised learning consists of some classifiers like Decision tree classifiers, Liner classifiers, Rule-based classifiers and Probabilistic classifiers. SVM and Neural Networks are the linear classifiers whereas Naïve Bayes, Bayesian Network, and Maximum Entropy is Probabilistic classifiers. Lexicon Based approaches are classified into Dictionary based and Corpus-based methods. The corpus-based approach further divided into the statistical and semantic approach. The analysis of sentiments can be Document based where the opinion in the entire document is summarized or classified as positive, negative or neutral, other is Sentence based where various sentences in the text are classified or analyzed and Lastly, Sentiment Analysis can also be Phrase level based where the polarity is used, and classification is done. Various applications uses the Sentiment Analysis like in consumer market for product reviews, social media for finding general opinions about recent topics, to know the consumer trends etc. The remainder of this paper is organized as follows: Section 2 gives an overview of related work already done on Sentiment Analysis. Section 3 includes result and a comparative table of existing systems. Section 4 concludes survey. II. RELATED WORK Sentiment Analysis has been done using different techniques or methods. Some works extract the meaning of the text, document or sentence level while others obtain connections between users to assign sentiment polarity for sentiment analysis. In particular, sentiment analysis of tweets has been done not only using approaches based on text, i.e. lexicon based classifiers, but also by combining Natural Language Processing and Machine Learning techniques. Tang and Qin in [1] proposed A Joint Segmentation and Classification Framework for Sentence Level Classification. Here a candidate generation model is developed to produce segmentation candidates of a sentence and a segmentation ranking model used to score the usefulness of segmentation candidates for sentiment classification and also classification model for predicting the sentiment polarity. The experiments are performed on two datasets: a tweet dataset and a movie review dataset. The effectiveness of this approach is verified by applying it to these two datasets. Neethu and Rajasri in [2], analyzed the Twitter posts about electronic products like mobiles, laptops, etc. using Machine Learning approach. A new feature vector is developed for classifying the tweets as positive, negative and extracts peoples’ opinion about products. Classifiers like Nave Bayes, SVM, Maximum Entropy and Ensemble classifiers are used for classification. All these different classifiers have the same accuracy for the new feature vector. This feature vector performs well for electronic products. Sentiment Analysis of Twitter has become one of the most important research Topic. Lek and Danny in [3] proposed an aspect-based sentiment classification approach to analyzing opinions for tweets. Different experiments are performed on the aspect-based sentiment classifier, to improve the performance of existing classifiers. The experimental results show that aspect level classifier approach outperforms existing methods. From experimental results, it suggests that a layered classification approach that uses the aspect-based classifier as the first layer classification and the tweet-level classifier as the second layer classification is more efficient than a classifier trained using target-dependent features. Also, this approach consistently improves the performance of existing sentiment classifiers. Xia, Xu in [4] has proposed a model called Dual Sentiment Analysis (DSA). By creating sentiment reversed review for each training and test review, data expansion
- 2. A Survey on Sentiment Analysis and Opinion Mining (IJSRD/Vol. 3/Issue 10/2015/235) All rights reserved by www.ijsrd.com 1030 technique is proposed. Also, DSA framework is developed for sentiment classification. Multi-domain datasets are used for the experiment. Classification accuracy is compared with Naïve Bayes, Linear SVM, etc. Topics in Twitter are more diverse and always dedicated themselves to a particular domain or subject. Liu, Li, and Cheng in [5], proposed an adaptive multiclass SVM model that transfers an initial common sentiment classifier to a topic adaptive. An iterative algorithm is proposed which is followed by three steps: optimization, unlabeled data selection and adaptive feature expansion steps. Comparison is done with the well-Known supervised sentiment classifiers like SVM, Decision Tree, Random Forest etc. and semi-supervised approaches. Accuracy of algorithm increased averagely 74% on the 6 topics from public tweet corpus. There are different methods to analyze information or text from particular document. But to extract and analyze the valuable information from text messages from social sites, Batool, Khattak in [6] proposed the Precise tweet classification technique. To find keywords from messages, feature extraction is used. After that semantic based filtering of data is done to categorize the data into specific category. By doing filtering on data, the information gain is increased i.e. knowledge enhancer is used to avoid information loss. To identify the interest of people’s or opinion towards social events, Zhou, Yong in [7] proposed a Tweet Sentiment Analysis Model (TSAM). Evaluation of TSAM is done on political dataset i.e. Australian Federal Election 2010. The proposed model provides fast and less expensive alternative approach over traditional polls for mining public views. The drawback of the approach is POS processing is not done. Also, Emotion Analysis is not included during Sentiment Analysis. To analyze the data, one can use Support Vector Machine (SVM) which is supervised learning method, but Basari, Hussain in [8] proposed a hybrid method of SVM and Particle Swarm Optimization (PSO). To solve the problem of dual optimization, hybrid method is proposed. For extraction of more features, from prepared datasets, PSO is used. The accuracy of SVM is affected by the hybridization of SVM-PSO and which is 77%. The other variation of SVM can be used for improving the accuracy. To perform Sentiment Analysis on News articles which gives fine grained analysis, Raina in [9] used Sentic Computing approach. For Concept-Level analysis, SenticNet; a publicly available resource is used. The performance of engine is measured by evaluation parameters like accuracy, precision, recall and F-measure. The proposed approach gives an accuracy 71.2%. This Method is only feasible for news articles i.e. it should be commercially viable. The drawback found is verb + noun pair is not included in Semantic Parser. So Semantic Parser can be improved by extracting verb + noun pair. III. RESULT AND COMPARISON By above related work, comparison of different mechanism used for the sentiment classification is done. The Overview of various existing techniques, methods and algorithms are described in the following table. The Paper title, Datasets used, advantages, disadvantages and accuracy are included in the table. Paper title Datasets Used Techniques/ Algorithms Advantages Drawbacks Accuracy 1) A joint segmentation and classification framework for sentence level sentiment classification. Tweet dataset and movie review dataset. Training algorithm and prediction algorithm. -A joint modelling framework outperforms over pipeline method in various experimental settings. -In joint framework static datasets are used i.e. publicly available dataset. 81%. 2) Sentiment analysis in twitter using machine learning techniques Tweets of electronic products. Naïve Bayes , SVM , Max entropy. -Simpler & efficient by using machine learning techniques. -Performs well for electronic products. - Domain dependent i.e. only feasible for electronic products. NB-89.5% SVM, Max Entropy- 90% 3) Aspect based twitter sentiment Classification. Stanford Twitter Sentiment (STS) consists of 1.6 million tweets. Aspect based sentiment classification. Aspect based classifier consistently increasing the performance of classifier. -In the aspect based classification, Contextual information is not considered. 84% 4) Dual Sentiment Analysis considering two sides of one review. Chinese Hotel, Notebook, 4 English Dataset. Dual Sentiment Analysis (DSA) model. -Effective for polarity classification. - Strengthen the DSA algorithm by data expansion technique. - In the Dual Sentiment Analysis, Complex polarity shift patterns are not used. 73% 5) Adaptive Co- training SVM for Sentiment Classification of tweets. Sanders Twitter Sentiment Corpus & presidential debate corpus. Multiclass SVM, Adaptive co- training algorithm. Adaptive co-training SVM expands topic adaptive feature. -Less features are extracted. -Timeline visualization of result is not done. 74%
- 3. A Survey on Sentiment Analysis and Opinion Mining (IJSRD/Vol. 3/Issue 10/2015/235) All rights reserved by www.ijsrd.com 1031 6) Precise Tweet Classification and Sentiment Analysis. Diabetic relevant Dataset. Precise Tweet Classification Model. -This is a method to collect more precise information about specific Topic. - 72.8% 7) Sentiment Analysis on Tweets for Social Events. Australian Federal Election 2010. Tweet Sentiment Analysis Model (TSAM). Knowledge enhancer is used to avoid information loss. -POS processing is not done. -Also, Emotion Analysis is not included during Sentiment Analysis. - 8) Opinion Mining of Movie Review using Hybrid Method of Support Vector Machine and Particle Swarm Optimization. Movie review dataset. Support Vector Machine and Particle Swarm Optimization (SVM-PSO). -Method solves the problem of dual optimization. -Hybridization gives better result. -To increase the accuracy, SVM with other optimization methods can be used. 77% 9) Sentiment Analysis in News Articles Using Sentic Computing. Multi- Perspective Question Answering (MPQA) corpus, a large corpus of news articles. News-focused opinion-mining engine, sentic computing- based Approach. -Method is reliable For identifying neutral sentences. -only feasible for news articles i.e. it should be commercially viable. -verb + noun pair is not included in Semantic Parser. 71.2% Table 1: Overview of various existing techniques and algorithms IV. CONCLUSION Internet and Web technologies are continuously growing and expanding, so the space and scope in the area of information retrieval is also expanding. For that, Sentiment Analysis and Opinion Mining is considered as an interesting area of research due to its many applications in society. Above survey explore different techniques, also these Opinion Mining techniques suffers from a number of problems, such as accuracy, scalability, quality, standard of data, domain dependency, etc. These problems have to be tackled separately and those solutions can be used to improve the methods to do sentiment analysis and classification. REFERENCES [1] Duyu Tang, Bing Qin, Furu Wei, Li Dong, Ting Liu, and Ming Zhou, “A joint Segmentation and Classification Framework For Sentence Level Classification”, IEEE/ACM TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING, VOL. 23, NO. 11, NOVEMBER 2015. [2] Neethu M S, Rajasree R, “Sentiment Analysis in Twitter using Machine Learning Techniques”, 4th ICCCNT 2013 ,July 4 - 6, 2013, Tiruchengode, India. [3] Hsiang Hui Lek and Danny C.C. Poo, “Aspect-based Twitter Sentiment Classification”, 2013 IEEE 25th International Conference on Tools with Artificial Intelligence. [4] Rui Xia, Feng Xu, Chengqing Zong, Qianmu Li, Yong Qi, and Tao Li, “Dual Sentiment Analysis: Considering Two Sides of One Review”, IEEE Transactions on Knowledge and Data Engineering, MANUSCRIPT ID, 1041-4347 (c) 2015 IEEE. [5] Shenghua Liu, Fuxin Li, Fangtao Li, Xueqi Cheng, Huawei Shen, “Adaptive Co-Training SVM for Sentiment Classification on Tweets”, Copyright 2013 ACM 978-1-4503-2263-8/13/10. [6] Rabia Batool, Asad Masood Khattak, Jahanzeb Maqbool and Sungyoung Lee, “Precise Tweet Classification and Sentiment Analysis”, 978-1-4799- 0174-6/13/$31.00 ©2013 IEEE. [7] Xujuan Zhou, Xiaohui Tao, Jianming Yong, ” Sentiment Analysis on Tweets for Social Events”,2013 IEEE 17th International Conference on Computer Supported Cooperative Work in Design, 978- 1-4673-6085-2/13/$31.00 ©2013 IEEE. [8] Abd. Samad Hasan Basaria,*, Burairah Hussina, I. Gede Pramudya Anantaa, Junta Zeniarjab, “Opinion Mining of Movie Review using Hybrid Method of Support Vector Machine and Particle Swarm Optimization”, 1877-7058 © 2013 The Authors. Published by Elsevier Ltd. [9] Prashant Raina, “Sentiment Analysis in News Articles Using Sentic Computing”, 2013 IEEE 13th International Conference on Data Mining Workshops, © 2013 IEEE.