Anda di halaman 1dari 5

IJIRST –International Journal for Innovative Research in Science & Technology| Volume 3 | Issue 12 | May 2017

ISSN (online): 2349-6010

A Study & Comparison on Sentiment Analysis for

the Products Available in E- Commerce
S. Muthukumaran Dr. P.Suresh
Research Scholar Research Supervisor & Head of Department
Research & Development Centre Department of Computer Science & Engineering
Bharathiar Univesity, Coimbatore, India Salem Sowdeswari College, Salem, Tamilnadu, India

This paper explains different methods for sentiment analysis and showcases an efficient methodology. It also highlights the
importance the product reviews are of utmost importance for the buyers to decide based on their concerns regarding product's
various aspects for example a monitor, processor speed, memory etc. Hence this sentiment analysis of product review provides
nearly accurate statistics regarding a product, providing an ease to the customers for analyzing the product and zero down his/her
search for an online product. The key focus here is efficient feature extraction, polarity classification thereby summarizing
positive and negative or neutral polarity. The proposed work is able to collect information from various sites and perform a
sentiment analysis of a user reviews based on that information to rank a product. Also these reviews suffer from spammed
reviews from unauthenticated users. So to avoid this confusion and make this review system more transparent and user friendly
we propose a technique to extract feature based opinion from a diverse pool of reviews and processing it further to segregate it
with respect to the aspects.
Keywords: Sentiment Analysis, Python Language, Products, Opinion Mining, Natural language processing


Sentiment analysis (also known as opinion mining) refers to the use of natural language processing, text analysis and
computational linguistics to identify and extract subjective information in source materials. Sentiment analysis is widely applied
to reviews and social media for a variety of applications, ranging from marketing to customer service. The aim of Sentiment
analysis or Opinion mining is to determine the attitude of a speaker or a writer with respect to some topic or the overall
contextual polarity of a document. Sentiment analysis can help in defining the features of different web services, technologies
and social sites. The information available on the blogs, social networks, and forums provide many opportunities to the
developers this Sentiment analysis used widely in the industrial field. It has increased much consideration as of late and studies
individual’s emotions towards certain substances. Sentiment analysis deal with many challenges, which describes an element
considered in this social network areas. It is not a cup of tea. For example sentiments about any product can be determined by
positive and negative opinions of people on the product. This technique implements various algorithms to analyze the corpus of
data and make sense out of it. This technique helps to identify the orientation of a sentence thereby recognizing the element of
positivity or negativity in it. Automated opinion mining can be implemented through a machine learning based approach.
Opinion mining uses natural language processing to extract the subjective information from the data. Document level tasks
mainly help in segregating the overall document into either subjective document or objective document. Further it can be
distinguished into positive, negative or neutral. It can also help separate the spam from the non-spam.
The sentence level opinion mining is performed on the sentences which can help group certain sentences to summarize the
opinion and also it can help identify comparative sentences to rank them accordingly. Phrase level deals with the aspects and is
known as aspect based opinion mining. This helps to identify the reviewer’s sentiment about specific aspects of the product. This
level does the finer-grained analysis of the opinions. According to the reviews by experts, one can find out the quality of
technique. The analysis of sentiment work up with few difficulties: Among other things, it must be resolved whether document
or segment thereof is subjective or objective and whether the sentiment communicated is positive or negative. The sentiment of
the content may be vital in fewer applications:
 Customer, summarizing customer, software and product reviews.
 Classifying sites posts and comments.
The Sentiment analysis used ideally in the industrial field, in this sentiment analysis aims to determine the feelings of speaker
and writer with respect to the same document. The polarity of the document can be checked with different methods and give
some range to a word in order to determine polarity, the sentence that is positively considered as greatest polarity as compared to
the negative words. In the present work discovers the reviews of a specialized article. Determine the positive and negative survey
of individual of these articles. According to the reviews of the expert define how much that article technically sound or not.

All rights reserved by 191

A Study & Comparison on Sentiment Analysis for the Products Available in E- Commerce
(IJIRST/ Volume 3 / Issue 12/ 029)


A simple unsupervised learning algorithm for classifying a review as recommended or not recommended. Average Semantic
orientation of the phrases in the review that contain adjectives or adverbs is used for classifying the reviews. This paper works on
document level Sentiment Analysis. Point wise Mutual Information (PMI) is used calculates semantic orientation of phrase and
the word. They took reviews from Opinions for different domains (Automobiles, Banks, Movies, and Travel Destinations). They
got accuracy ranges from 84% for automobile reviews to 66% for movie reviews.
In they have developed an algorithm for predicting semantic orientation. Algorithm designed for isolated adjectives, rather
than phrases containing adjectives or adverbs. They used four step supervised learning algorithm to infer the semantic orientation
of adjectives from constraints on conjunctions. In that they got accuracy for classification of adjectives ranging from 78 % to 92
% depending on amount of training data. In they developed such system that generates sentiment timelines. It tracks online
discussions on movies and generate plot which contains number of positive sentiment and negative sentiment messages over
time. They used specific domain lexicons for movies. It is used instead of a hand-built lexicon. This work is used in automatic
review rating, tracking advertising campaigns, tracking public opinion for politicians, tracking financial opinions by stock
traders, tracking entertainment and technology trends by trend analyzers. In it is concerned with subjectivity tagging. They
evaluated objectively present factual information. This paper identifies strong clues of subjectivity using the results of a method
for clustering words according to distributional similarity. In 10-fold cross validation results, features based on both similarity
clusters and the lexical semantic features are shown to have higher precision than features based on each alone.
In classify a document's polarity on a multi-way scale and expanded the task of classifying a movie review as either positive or
negative to predicting star ratings on either 3 or 4 star scale. They checked human performance at the task. Applied algorithm is
Meta algorithm, Based on a metric labeling. This Meta algorithm can give best performance over both multi-class and regression
versions of SVMs when we employ a novel similarity measure appropriate to the problem. They used movie review Dataset. The
opinionated reviews also contain other information that can be used to ascertain the sentiment about a product. Venkata Rajeev P
et al uses the reviews from and proposes the combination of four parameters: star ratings of the product, the polarity
of the review, age of review and helpfulness score, for determining the opinion of a product. The task of mining the features is of
particular importance and many methods are suggested for it. Weishu Hu etal. Divides the opinion analysis tasks into three steps:
identifying the opinion sentences and their polarity, mining the features that are commented upon by customers, and removing
incorrect features.
The primary focus of product review system is identifying the adjective word in a sentence and identifying the sentiment
behind it. Yan Luo et al. suggests the final sentiment score of the review to be the cumulative sentiment score of all the
adjectives in that review D V Nagarjuna Devi et al. Proposes a system that uses a supervised classification approach called as
support vector machine. This paper claims that the proposed classifier approach gives out the best result. It also identifies various
challenges in sentiment analysis like sarcasm and conditional sentences, grammatical errors, spam detection and anaphora
resolution. Sentence level classification is done on input data which is further classified according to the subjectivity/objectivity.
Further aspect extraction is done using Senti Word Net. This is then further fed to SVM classifier to find the overall opinion.


Opinion Mining and Functions

Opinion retrieval is a document retrieving and ranking process. A relevant document must be relevant to the query and contain
opinions toward the query. Opinion polarity classification is an extension of opinion retrieval. It classifies the retrieved document
as positive, negative or mixed, according to the overall polarity of the query relevant opinions in the document. It proposes
several new techniques that help improve the effectiveness of an existing opinion retrieval system; which presents a novel two-
stage model to solve the opinion polarity classification problem. In this model, every query relevant opinionated sentence in a
document retrieved by our opinion retrieval system is classified as positive or negative respectively by a machine learning
technique which analysis the comparison on data report. Then a second classifier determines the overall opinion polarity of the
document. Experimental results show that both the opinion retrieval system with the proposed opinion retrieval techniques and
the polarity classification model outperformed the best reported systems respectively.
Existing Phase
In order to achieve maximum accuracy in feature extraction process, the noises present in the user generated content should be
eliminated. The noises are usually in the form of spelling mistakes, grammar mistakes, mistakes in punctuation, incorrect
capitalization, and usage non- dictionary words such as abbreviations or acronyms of common terms and so on. The main reason
for this is these reviews are mostly written by non-experts and in short informal texts. After downloading the datasets from
internet, the proposed system cleaned the documents by removing the html tags present in the document and it correct spelling
errors. The texts are tokenized into tokens and the stop-words are detected and removed. Since words like preposition, digits,
articles and proper nouns like name of cell phone etc. are considered as valueless in the sentiment analysis, hence these words
are included in the stop word list. The sentences generated in this pre-processing can be parsed automatically by any linguistic

All rights reserved by 192

A Study & Comparison on Sentiment Analysis for the Products Available in E- Commerce
(IJIRST/ Volume 3 / Issue 12/ 029)

parser. The proposed system used Stanford Linguistic parser for POS tagging of each word present in the sentences. POS tagger
parses each sentences and tags each term with its part of speech.


Analytical status
Following are the approaches for machine learning:
1) Machine learning based approach includes:
1) supervised learning
2) unsupervised learning
3) semi-supervised learning
4) lexicon based.
2) Using Semantic Orientation scheme of extracting relevant n-grams of the text and then labeling them either as positive or
negative and consequentially the document.
3) Senti Word Net approach -It is based on analysis of the glosses associated to synsets and on the use of the resulting vectorial
term representations of semi-supervised synset classification. Senti Word Net is computationally a favorable algorithm but
achieves relatively lower accuracy.
The method discussed in this paper is based on semi supervised learning which uses Word Net as lexicon based data
dictionary to convert features into target words.

Fig. 1: Public Review on Products for E-Commerce

Fig. 2: Country wise Public Review on Products for E-Commerce

The documents are cleaned by removing the html tags present in the document and by correcting spelling errors. The texts are
then tokenized into tokens and the stop-words are detected and removed. Stanford Linguistic parser is used for POS tagging of
each term. By applying the six rules, the features, opinions and modifiers are extracted. By applying a threshold frequency limit
of 3, the irrelevant terms are filtered out.
The dataset used for this project is the Flip kart Reviews Database. The reviews in the dataset are consists of the attributes
such as: Reviewer ID, Product ID, Review Text, Rating and time of the review. The main source of data used is the product
reviews from Amazon. The reviews for a few popular phones have been obtained by building a web crawler. The web crawler
has been written in Python using a scraping library called Beautiful Soup. Along with the review text, some additional data

All rights reserved by 193

A Study & Comparison on Sentiment Analysis for the Products Available in E- Commerce
(IJIRST/ Volume 3 / Issue 12/ 029)

related to the reviews such as reviewer name, review date, overall rating and comments were also obtained. The crawler is called
periodically to get the most up-to-date reviews.
Python Language Techniques
Python is, or can be used as the scripting language in these software products:
 Abaqus (Finite Element Software)
 ADvantage Framework.
 Amarok.
 ArcGIS, a prominent GIS platform, allows extensive modeling using Python.
It is a general-purpose programming language typically used for web development and SQ Lite is one free lightweight
database commonly used by Python programmers to store data.


This paper shows that, the system performs very well in sentiment classification of user reviews with high accuracy. The
implemented fuzzy functions to emulate the effect of various linguistic hedges such as dilators, concentrator and negation on
opinionated phrases help the system to achieve more accuracy in sentiment classification and summarization of users’ reviews in
various aspects and various countries. As future work of this research, we can refine rule set to extract more dependency
relations from datasets and that will help to improve the precision and recall values of the system by defining algorithms. From
the analysis of review documents, it fails the system from defining correct dependency relations between word pairs and
comparison results. If the system able to correct all the spelling and grammatical errors present in the review documents in the
pre-processing step itself that will definitely improve the recall value of the system.

[1] Venkata Rajeev P, Smrithi Rekha V, “Recommending Products to Customers using Opinion Mining of Online Product Reviews and Features”, 2015
International Conference on Circuit, Power and Computing Technologies [ICCPCT]
[2] Weishu Hu, Zhiguo Gong, Jingzhi Guo, “Mining Product Features from Online Reviews”, IEEE International Conference on E-Business Engineering.
[3] Yan Luo,Wei Huang, “Product Review Information Extraction Based on Adjective Opinion Words”, 2011 Fourth International Joint Conference on
Computational Sciences and Optimization.
[4] A.S.Syed Navaz, A.S.Syed Fiaz, C.Prabhadevi, V.Sangeetha & S.Gopalakrishnan “Human Resource Management System” Jan – Feb 2013, International
Organization of Scientific Research Journal of Computer Engineering, Vol 8, Issue 4, pp. 62-71.
[5] Pang, Bo; Lee, Lillian (2005). "Seeing stars: Exploiting class relationships for sentiment categorization with respect to rating scales". Proceedings of the
Association for Computational Linguistics (ACL). pp. 115–124.
[6] A.S.Syed Navaz, M.Ravi & T.Prabhu, “Preventing Disclosure of Sensitive Knowledge by Hiding Inference” February 2013, International Journal of
Computer Applications, Vol 63 – No 1. pp. 32-38.
[7] A.S.Syed Navaz , T.Dhevisri & Pratap Mazumder “Face Recognition Using Principal Component Analysis and Neural Networks“ March -2013,
International Journal of Computer Networking, Wireless and Mobile Communications. Vol No – 3, Issue No - 1, pp. 245-256.
[8] D V Nagarjuna Devi, Chinta Kishore Kumar,Siriki Prasad, “A Feature Based Approach for Sentiment Analysis by Using Support Vector Machine”, 2016
IEEE 6th International Conference on Advanced Computing.
[9] Hatzivassiloglou, V., & McKeown, K.R. 1997. Predicting the semantic orientation of adjectives. Proceedings of the 35th Annual Meeting of the ACL and
the 8th Conference of the European Chapter of the ACL (pp. 174-181). New Brunswick, NJ: ACL.
[10] A.S.Syed Navaz, P.Jayalakshmi, N.Asha. “Optimization of Real-Time Video Over 3G Wireless Networks” September – 2015, International Journal of
Applied Engineering Research, Vol No - 10, Issue No - 18, pp. 39724 – 39730.
[11] Tong, R.M. 2001. An operational system for detecting and tracking opinions in on-line discussions. Working Notes of the ACM SIGIR 2001 Workshop on
Operational Text Classification (pp. 1-6). New York, NY: ACM.
[12] Wiebe, J.M. 2000. Learning subjective adjectives from corpora. Proceedings of the 17th National Conference on Artificial Intelligence. Menlo Park, CA:
AAAI Press.
[13] Yang Liu, Xiangji Huang, Aijun An, Xiaohui Yu (2007). “ARSA: A Sentiment-Aware Model for Predicting Sales Performance Using Blogs” SIGIR‟07,
July 23–27, 2007, Amsterdam, The Netherlands.
[14] R Core Team (2015). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL
[15] A.S.Syed Navaz, H.Iyyappa Narayanan & R.Vinoth.” Security Protocol Review Method Analyzer (SPRMAN)”, August – 2013, International Journal of
Advanced Studies in Computers, Science and Engineering, Vol No – 2, Issue No – 4, pp. 53-58.
[16] A.S.Syed Navaz & K.Girija “Hacking and Defending in Wireless Networks” Journal of Nano Science and Nano Technology, February 2014, Volume- 2,
Issue – 3, pp - 353-356.
[17] A.S.Syed Navaz & R.Barathiraja “Security Aspects of Mobile IP” Journal of Nano Science and Nano Technology (International), February 2014, Volume-
2, Issue – 3, pp - 237-240.
[18] A. Kennedy and D. Inkpen, “Sentiment classification of movie reviews using contextual valence shifters”, Computational Intelligence, vol. 22, no. 2, pp.
110–125, 2006.
[19] A.S.Syed Navaz, G.M. Kadhar Nawaz & B.Karthick.” Probabilistic Approach to Locate a Mobile Node in Wireless Domain”, April – 2014, International
Journal of Computer Engineering and Applications, Vol No – 6, Issue No – 1, pp.41-49.
[20] Cataldo Musto, Giovanni Semeraro, Marco Polignano."A comparison of Lexicon-based approaches for Sentiment Analysis of microblog posts".8th
International Workshop on Information Filtering and Retrieval Pisa (Italy) December 10, 2014.
[21] M. Kristina Toutanova, Dan Klein and Y. Singer. “Feature-rich partof-speech tagging with a cyclic dependency network”. In HLT-NAACL, pages 252–
259. ACM, 2003.
[22] Ms.K.Mouthami, Ms.K.Nirmala Devi, Dr.V.Murali Bhaskaran, “Sentiment Analysis and Classification Based On Textual Reviews”, Information
Communication and Embedded Systems (ICICES), 2013 International Conference on 21-22 Feb. 2013Page(s): 271 – 276.

All rights reserved by 194

A Study & Comparison on Sentiment Analysis for the Products Available in E- Commerce
(IJIRST/ Volume 3 / Issue 12/ 029)

[23] L. Polanyi and A. Zaenen, “Contextual valence shifters”, in Computing Attitude and Affect in Text: Theory and Applications, vol. 20 of The Information
Retrieval Series, pp. 1–10, 2006.
[24] A.S.Syed Navaz & Dr.G.M. Kadhar Nawaz & A.S.Syed Fiaz “Slot Assignment Using FSA and DSA Algorithm in Wireless Sensor Network” October –
2014, Australian Journal of Basic and Applied Sciences, Vol No –8, Issue No –16, pp.11-17.
[25] A.S.Syed Navaz, J.Antony Daniel Rex, P.Anjala Mary. “An Efficient Intrusion Detection Scheme for Mitigating Nodes Using Data Aggregation in Delay
Tolerant Network” September – 2015, International Journal of Scientific & Engineering Research, Vol No - 6, Issue No - 9, pp. 421 – 428.
[26] S.Jensy Mary, A.S Syed Navaz & J.Antony Daniel Rex, “QA Generation Using Multimedia Based Harvesting Web Information” November – 2015,
International Journal of Innovative Research in Computer and Communication Engineering, Vol No - 3, Issue No - 11, pp.10381-10386.
[27] A.S Syed Navaz & K.Durairaj “Signature Authentication Using Biometric Methods” January – 2016, International Journal of Science and Research, Vol
No - 5, Issue No - 1, pp.1581-1584.
[28] S. Nadali, M. A. A. Murad, and R. A. Kadir, “Sentiment classification of customer reviews based on fuzzy logic”, in Proceedings of the International
Symposium on Information Technology (ITSim’ 10), pp. 1037–1044, mys, June 2010.
[29] M. Hu and B. Liu, "Mining and summarizing customer reviews", Proceedings of the tenth ACM international conference on Knowledge discovery and data
mining, Seattle, 2004, pp. 168-177.
[30] B. Pang, L. Lee, and S. Vaithyanathan, “Thumbs up?: sentiment classification using machine learning techniques”, Proceedings of the ACL-02 conference
on Empirical methods in natural language processing, vol.10, 2002, pp. 79-86.
[31] K. Dave, S. Lawrence, and D. M. Pennock, “Mining the peanut gallery: Opinion extraction and semantic classification of product reviews”, Proceedings of
WWW, 2003, pp. 519–528.
[32] Python (programming language) -
[33] ”Stanford Core NLP Toolkit” Available:
[34] Inferring networks of substitutable and complementary products J. McAuley, R. Pandey, J. Leskovec Knowledge Discovery and Data Mining, 2015.
[35] V.K. Singh, R. Piryani, A. Uddin P. Waila, Marisha “Sentiment Analysis of Textual Reviews ,Evaluating Machine Learning, Unsupervised and
SentiWordNet Approaches” proceeding in IEEE 2013 5th International Conference on Knowledge and Smart Technology (KST).
[36] ZohrehMadhoushi, Abdul RazakHamdan,SuhailaZainudin“Sentiment Analysis Techniques in Recent Works” proceeding in Science and Information
Conference 2015.
[37] A.S.Syed Fiaz, N.Asha, D.Sumathi & A.S.Syed Navaz “Data Visualization: Enhancing Big Data More Adaptable and Valuable” February – 2016,
International Journal of Applied Engineering Research, Vol No - 11, Issue No - 4, pp.–2801-2804.
[38] A.S.Syed Navaz & Dr.G.M. Kadhar Nawaz “Flow Based Layer Selection Algorithm for Data Collection in Tree Structure Wireless Sensor Networks”
March – 2016, International Journal of Applied Engineering Research, Vol No - 11, Issue No - 5, pp.–3359-3363.
[39] A.D. Vo and C.Y. Ock, “Sentiment classification: a combination of PMI, SentiWordNet and fuzzy function”, in Proceedings of the 4th International
Conference on Computational Collective Intelligence Technologies and Applications (ICCCI ’12), vol. 7654, part 2 of Lecture Notes in Computer Science,
pp. 373–382, 2012.
[40] A.S.Syed Navaz & Dr.G.M. Kadhar Nawaz “Layer Orient Time Domain Density Estimation Technique Based Channel Assignment in Tree Structure
Wireless Sensor Networks for Fast Data Collection” June - 2016, International Journal of Engineering and Technology, Vol No - 8, Issue No - 3, pp.–1506-
[41] M.Ravi & A.S.Syed Navaz "Rough Set Based Grid Computing Service in Wireless Network" November - 2016, International Research Journal of
Engineering and Technology, Vol No - 3, Issue No - 11, pp.1122– 1126.
[42] A.S.Syed Navaz, N.Asha & D.Sumathi “Energy Efficient Consumption for Quality Based Sleep Scheduling in Wireless Sensor Networks” March - 2017,
ARPN Journal of Engineering and Applied Sciences, Vol No - 12, Issue No - 5, pp.–1494-1498.
[43] Kerstin Denecke “Using SentiWordNet for Multilingual Sentiment Analysis” 2008 IEEE.
[44] A.S.Syed Navaz, S.Gopalakrishnan & R.Meena “Anomaly Detections in Internet Using Empirical Measures” February 2013, International Journal of
Innovative Technology and Exploring Engineering, Vol 2 – Issue 3. pp. 58-61.
[45] Zhang Yunliang, Zhu Lijun, QiaoXiaodong, Zhang Quan “Flexible KNN Algorithm for Text Categorization by Authorship based on Features of Lingual
Conceptual Expression” 2008 IEEE ,World Congress on Computer Science and Information Engineering.
[46] Federica Bision, Paolo Gastaldo, Chiara Peretti, Rodolfo Zunino and Erik Cambria “Data Intensive Review Mining for Sentiment Classification across
Heterogeneous Domains” 2013 IEEE at ACM International Conference on Adavances in Social Networks Analysis and Mining.

All rights reserved by 195