Sentimental Analysis and Opinion Mining on Online Customer Review

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 09 Issue: 10 | Oct 2022 www.irjet.net p-ISSN: 2395-0072

Sentimental Analysis and Opinion Mining on Online Customer Review

Anushree Raji , Joylin Denita Dsouza2, Shabarish S.K3

1 Assistant Professor- IT Department, AIMIT, Mangaluru, anushreeraj@staloysius.ac.in

2 MCA Student, AIMIT, Mangaluru, 2117026joylin@staloysius.ac.in

3 MCA Student, AIMIT, Mangaluru, 2117103shabarish@staloysius.ac.in ***

Abstract: Customer Opinions play a very crucial role in daily life. The views of other people are taken into account when we must make a decision. Today, a large number of internet users share their ideas about a variety of products on blogs, review websites, and social networking sites. The number of individuals buying things from the web is growing, and there are an ever-increasing number of results kept there. As a result, the number of user evaluations or postings is growing daily. Business organizations andcorporateorganizationsare constantly interested in hearing what consumers or individuals have to say about their goods, services, and support. Today, if someone needs to purchase an online item, he/she can first view other buyer’s reviews on the product site and take the right decision accordingly. There are many opinions on the web about a product so takingdecisionsmight be difficult. Therefore, opinion mining is used for classifying the reviews according to their polarity. The method of extracting opinions from reviews is known as opinion mining (OM). Any user who wants to choose a product or business should consider consumer reviews when doing so. Sentiment analysis is another name for opinion mining. Our main goal is to develop a system for analyzing opinions, which implies judging various consumer products.

Key words: Opinion Mining, Sentiment Classification, Online Reviews, Data Mining, Web Mining, Sentiments analysis (SA).

1. INTRODUCTION

The current era is characterized by the presence of Ecommerce retailers everywhere around us. Almost all companyphasesareE-commercestores.Withwidespread Internet connection and knowledge of the method, the Ecommerce business has soared to new heights in recent years. There are several criteria that contribute to an Ecommercestore'ssuccessandcredibility.ProductReviews, ontheotherhand,areasignificantcomponentinraisingthe reputation, standard, and assessment of an Ecommerce company [1]. Product Reviews are one of the most significanttoolsaccessibletoanEcommerceshop,namely CustomerFeedback.

OnecriticaldutyfortheEcommercecompanyistokeepup itsreputationinthewebindustry.Naturally,ittakesalotof efforttogetthatreputation,butitcostsverylittletoloseit: ProductReviewsarethemostidealapproachestokeepup their series of wins. Item reviews and criticisms have

modifiedtheenjoymentfortheinternetindustrybecausethe web has become so widespread. Product Reviews are the elements that determine the customer's trustworthy relationship with the business - they assist build dependabilityandtrustandexplainthepossiblebuyerthe item significantly more clearly and the viewpoints that distinguishitfromwhateverisleftofthethingsanywhere else.

AnEcommerceshopwithagoodcollectionofclientreviews for its products displays a good reputation among customers. Presently the opinions of others regarding a productinfluencepurchasingdecision[2].Forexample,the consumer will just purchase the item by reading the feedback written by customers. he obtains clear thinking about the determination and effectiveness of the organization's product specifics and intricacies to their products However, in real-life situations, half of The highlights that the producer gives about the item are not true. As a result, only authentic customers who use that productmayprovidespecificinformationabouttheproduct Now arrives the importance of reviews We are currently witnessing financial waste as a result of poor purchase decisions.

Thepresentationofsemanticanalysisonreviewstacklesthe aboveproblem.Users’premiumissteadyjustinbriefperiod. so, client subjects from surveys can be delegate for e.g. if there should arise an occurrence of a versatile phone, different individuals have distinctive ideas. a few people focusoncamera,wheresomefocusonbatteryreinforcement thuson.Theyallhavecustomizedterritoryofenthusiasmfor theitem.Theimportanceofsentimentanalysiscomeshere. Sentiment analysis otherwise known as opinion mining is the process of determining the emotional tone behind a series of words. Sentiment analysis is extremely useful in onlinee-commercesitestomonitorthereviewsitallowsus to gain an opinion about the product. Using sentiment analysis on product reviews helps us to extract the emotional tone towards the product. Through natural languageprocessingandmachinelearning.productreviews ine-commercesitesarewritteninnaturallanguagessuchas English.Thistechniqueisusedtofigureoutthesentimentor emotionassociatedwiththeunderlyingtext.So,ifyouhavea piece of text and you want to understand what kind of emotionitconveys,forexample,anger,love,hate,positive, negative,andsoonyoucanusethetechniquesentimental analysis

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 09 Issue: 10 | Oct 2022 www.irjet.net p-ISSN: 2395-0072

2. OBJECTIVES

Thepurposeofsentimentanalysisandopinionminingisto teachcomputerstounderstandandcommunicateemotion. Thisworkfocusesonminingreviewsfromwebsitessuchas Amazon,whichallowuserstofreelypublishtheiropinions. Thereviewsareautomaticallyextractedfromthewebsite. Mostsemi-structuredonlinedatacontainawealthofhelpful information. With the increasing growth of ecommerce, more and more things are being sold online. People are increasinglydoingtheirpurchasingontheinternet.Notonly that,butconsumersfrequentlydiscusstheirproduct-related experiencesontheinternet.Theyutilizeblogs,Twitter,and othersuchwebsitestoexpressthemselvesandsharetheir experiences with others. Websites for social interactions havegrowninpopularity.Ithasbecomenormalpracticefor onlineretailerstoallowtheircustomerstoevaluateorvoice thoughtsaboutthethingsthattheyhavepurchasedinorder toimprovecustomerhappinessandshoppingexperience.

Asthenumberofonlinebuyersgrows,sodoesthenumber ofreviewsexpressontheinternet.Hundredsofthousandsof reviews have been written about some products. Understanding the consumers' perceptions of products is extremelybeneficialtobothmerchantsandcustomerswho want to purchase those products in the future. When the quantityofreviewsishuge,readingthemallonebyoneis inefficient. The review material also sometimes generates confusions. The majority of product reviews have a lot of longsentences.Fewofthemtrulyexpresstheiropinion.Asa result, it is more difficult to read and comprehend comments.

If someone reads only a few reviews and then makes a judgement,thedecisionmaybeprejudiced.Becauseofthese factors,havingabetterdataminingtechniquetominethese semi-structured product reviews is critical. Not only are product reviews significant, but so are reviews on certain places,sports,andmoviesiftheyareminedcorrectlytoget theirrealviewpoint.Manyresearchershavelookedintothis issue in recent years [3]. The field of study is known as opinionminingandsentimentanalysis.Thisresearcharea has two primary goals. They are (1) locating product features on which reviewers have remarked and (2) determiningwhethertheremarksarepositiveornegative. Both activities are extremely difficult, and several studies havebeenundertakenonthem.

3. LITERATURE REVIEW

Rajkumar et al. rendered ML approaches say Naïve Bayes (NB) and SVM for performing SA on reviews of a specific product.Inthoseapproaches,thedatasetwasgatheredasof Amazon, which comprised reviews regarding Laptops, Cameras, Mobiles, Tablets, video surveillance, and TVs. Subsequently, stemming, stop word removal, and also punctuation marks removal was executed and it was transmutedintoabagofwords.Thisdatasetwascontrasted

toopinionlexicons,thatis,4783negativeand2006positive words with sentiment scores intended for every sentence wereevaluated.Utilizingscoreanddisparatefeatures,the NBalongwithSVMwereemployedanddiverseaccurateness was computed. The ML approaches proffered the good outcomes to categorize product reviews. NB got 98.170% accuracyandSVMgot93.54%accuracyforCamera related Reviews.TheapproachutilizestheSVM,whichencompasses severalkeyparametersthatarerequiredtobesetproperly forattainingthebestclassificationoutcomes.Thus,theSVM rendersloweraccuracyinclassification[4].

SatuluriVanajaandMeenaBelwalrenderedanAspect-level SA,whichwasattained byIdentification,aggregation, and Classification.Thepre-processingincludesParts-of-Speech taggingtoeverywordineachsentence,extractingfrequently used words, removing stopping or unwanted words and adjective extraction from the sentences. The classification was executed utilizing ML algorithms, NB and SVM classification algorithm and the performance were contrasted centered on Recall, F1 measure, and Precision. The outcomes evinced that it attained more accurateness from the NB when weighed against the SVM. The SVM approachwasnotaptforlargedatasets[5].

Wei Zhang et al. propounded an emotion classification algorithm grounded on SVM as well as latent-SA (LSA). Primarily,PsychologyandNLPwereintegratedtodividethe emotions in the online reviews onto ‘4’ categories: a. happiness,b.hope,c.disgust,andd.anxiety.Subsequently, theLSAapproachwasutilizedforoptimizingthetextfeature extraction and employed the SVM as a classifier for amelioratingtheemotionalclassificationaccuratenessand computationalefficacy.Theexperientialoutcomesevinced that the model could effectually compute online reviews. ContextmeaningsofdatawithDLalgorithmswereutilized forcombiningthereviews’theme,sentimentclassification and product characteristics for further enhancing the multipleclassemotionaldetectionaccuracy.Theapproach employsonlyalessamountofdataforanalyzing,whichis notefficient[6].

BarkhaBansalandSangeetSrivastavarenderedaHybridized Attribute-centric Sentiment Classification (HABSC) for infusing domain-specific knowledge and collecting the implicit word relations. This approach found the utmost frequentbi-gramaswellastri-graminthecorpus,followed by POS tagging for retaining opinion words and aspect descriptions. Subsequently, it has deployed TFIDF for signifyingeverydocument,followedbyautomaticextraction of an optimal topic. All the adverbs and adjectives were labelled utilizing pre-existing lexicon and domain-related knowledge. The method’s efficacy was tested utilizing datasets. The outcomes evinced that the classification accuratenessofHABSCexceededfordisparatemethodsand also evinced less computational time as contrasted to distributedvectorizationframeworks.Theapproachwasnot

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 09 Issue: 10 | Oct 2022 www.irjet.net p-ISSN: 2395-0072

effective in detecting the attribution, and it has a high computationaltime[7].

ChonghuiGuoetal. examinedarankingapproachviaonline reviews grounded on diverse aspects of variant products thatintegratedthesubjectiveaswellasobjectivesentiment values. Primarily, the product’s sentiment value was evaluatedbyascertainingtheweightsofthoseaspectswith theLDAtopicdesign.Atthetimeofthisprocess,therealistic meaning of every single aspect was as well summarized. Subsequently, consumers’ personalized preferences were regarded whilst evaluating the total scores of variant products.Meanwhile,comparativesuperiority in-between every‘2’productsalsoaddedintofinalscores.Byutilizing the Page Rank algorithm, the attained final score of every product was evaluated as of the constructed graph. The outcome elucidated that whilst regarding only objective sentiment values of the product, the ranking outcome attained by thisapproachhada goodcorrelation with the primarysalesorders.ButthesystemusedtheLDA,which wassensitivetooverfit,andvalidationofLDAmodelswasat leastproblematic[8].

SumbalRiazetal.recommendedanapproachtermedtext mining for examining customer reviews to ascertain the customers’ opinions and executed the SA on the massive datasetofproduct(6sorts)reviewsprofferedbydisparate customers on the internet. In this approach, SA was employedatthephraselevelinsteadofdocument-levelfor computing every term’s SP. Then key graph keyword extractionwasusedaimedatextractingkeywordsasofeach documentwithhigh-frequencytermsandtheintensityofSP by gauging its strength was evaluated. The k-means clustering was utilized for grouping data on the base of sentimentstrengthvalue.Thosevalueswerecontrastedto thestarratingofthesamedataandtheexcellentandneutral sentiment toward products was examined. The approach usesclusteringwhichmaybringaboutoverclustering[9].

4. METHODOLOGY

This research paperwork is divided into modules. This researchstudyfocusesonAmazonproductreviewmining that follows the free review structure. There will be no constraintsontheuser'sabilitytowriteareview.Usersof the online shopping site Amazon are encouraged to leave productreviewsontheitemstheypurchase.Amazonusesa 1-to-5scaleforallproducts,regardlessofcategory,makingit difficult to assess the benefits and drawbacks of various elements of a product. It explains how to extract product attributes from opinion statements. It employs a SentiWordNet-basedalgorithmtodeterminethesentence's opinion.

Sentiment analysis

Sentimentanalysisistheautomatedmethodofdeducinga person'sfeelingsonaspecifictopicfromwrittenorspoken

language.Sentimentanalysisissometimesknownasopinion mining,anditisabranchofnaturallanguageprocessingthat extractshiddenopinionsintext.

In extracting an expression, there are three attributes to consider:

a) polarity- whatkindofpolaritythecustomerexpressesin hisreview;itcanbepositive,negative,orneutral.

b) subject- thethingbeingdiscussed.

c) opinion holder-thecustomerwhoexpressesanopinion aboutaproductthroughreviews.

Becauseofitsmultiplefunctionaluses,sentimentanalysisis currently a subject of extraordinary priority and advancement. Because the amount of freely and covertly accessible data on the Internet is always rising, a large numberofworksdescribingfeelingsareavailablethrough reviewsites,conversations,webjournals,andsocialmedia. Withtheassistanceofhypothesisexaminationframeworks, this unstructured data could be transformed into ordered knowledge on public sentiments regarding things, administrations, brands, legislative difficulties, or any subjectaboutwhichpeoplecanexpressopinions[10].This data can be used for commercial applications such as showcasing analysis, advertising, item surveys, net advertiserscore,iteminput,andclientmanagement.

Scope of Sentiment Analysis

Sentimentanalysiscanbeusedatvariouslevelsofdetail: 

Document level sentiment analysis obtains the sentimentofacompletedocumentorparagraph. 

Sentence level sentiment analysis obtains the sentimentofasinglesentence. 

Sub-sentence level sentimentanalysisobtainsthe sentimentofsub-expressionswithinasentence.

The kind of sentiment analysis

Therearedifferenttypesofsentimentanalysiswhereinthis systemweproposeacombinationoffine-grainedsentiment analysis, emotion detection, and aspect-based sentiment analysis.

Fine-grained Sentiment Analysis

Insteadof onlylookingatgenericviewpoints,weare now delving further into opinion mining. Instead of taking positive, neutral and negative opinions can considers the followingcategories: Verypositive 

Positive 

Neutral

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 09 Issue: 10 | Oct 2022 www.irjet.net p-ISSN: 2395-0072

Negative 

Verynegative

We can also utilize star representation, with 5 stars representingahighlygoodviewand1starrepresentinga verynegativeone.

Emotion detection

Emotion detection aims at detecting emotions like, happiness, frustration, anger, sadness etc. in the reviews. Justlikeminingtheopinionfromthereviewemotionsalso has its importance to form precise sentiment about a product.

Aspect-based Sentiment Analysis

Inthistypeofsentimentanalysis,wediscussnotonlythe sentiment of the review, but also which specific aspect or feature of the product we have an opinion about. For example,"thebatterylifeofthemobilephoneisinsufficient." Thesentenceexpressesabadviewaboutthemobilephone, morespecificallyaboutthebatterylife,whichisafeatureof thephone.

Working of sentiment analysis

Thereareseveralapproachesandalgorithmstoimplements sentimentanalysissystems,whichcanbeclassedas: 

Rule-based systems that execute sentiment analysis basedonasetofmanuallywrittenrules. 

Automatic systems thatlearnfromdatausingmachine learningtechniques. 

Hybrid systems that combine both rule based and automaticapproaches.

Opinion

The information in the text can be divided into two categories:factsandopinions.Whereasfactsareobjective statements,opinionsaresubjectiveexpressionsthatinclude userattitudesandfeelingsabouttheproduct[11].

Like other NLP challenges, sentiment analysis can be classifiedasaclassificationproblemwithtwosubproblems tosolve-Theyare:

Subjectivity classification-classifying the sentence into subjectiveorobjective

Polarity classification- classifyingthesentenceopinioninto positive,neutralandnegative

Inanopinion,theelementthecontentdiscussionsaboutcan beanitem,itssegments,itsaspects,itscharacteristics,orits

highlights.Itcouldlikewisebeanitem,anadministration,an individual, an association, an occasion, or a subject. As an example,takealookattheopinionbelow:

"Thebatterylifeofthismobilephoneisexcessivelyshort."A negativefeelingiscommunicatedaboutanelement(battery life)ofasubstance(mobilephone).

Direct vs. Comparative Opinions

There are two sorts of opinions: direct and comparative. Direct conclusions give a sentiment about a substance straightforwardly,forinstance:"Thesoundqualityofmobile phone A is poor." This direct opinion states a negative sentiment about mobile phone A. In comparative feelings, theopinioniscommunicatedbycontrastingasubstanceand another, for instance: "The sound quality of mobile A is betterthanthatofmobileB."

Among the different approaches to sentiment analysis accessible, (SA), only two major categories are prevalent. ThefirstcategorySA'sproblemsaresolvedbyimplementing themachine

method of learning Several techniques are used in this group. used in an attempt to extract significant traits that moreprovidepreciseinformationregardingthepolarityof sentiments

As the procedure progresses, the technique employed is regularly monitored. A carefully annotated corpus is required.Thesecondgroupemploysamethodthatismore linguisticallyinclined. referredtoasthelexicon-basedapproachaccordingtothe source. Theinvestigationbeginswithwordsorsentences. exhibitingsemanticpolarityqualities.

There are some a variety of machine learning (ML) methodologiesTheyhavecreatedforthegoalofcategorizing literatureaspositiveorbadorclassesthatarepositive.The approaches' performances: Naive Bayes (NB), Support Vector Machines (SVM), and Maximum Entropy (ME) and classification are used. really successful ID3 is one of the otherways.CentroidClassifier, Winnow Classifier, and K-Nearest Neighbor Mining ApproachforNeighborandAssociationRules.

NaïveBayes(NB)classificationmethodiscommonlyutilized forclassifyingtextdocuments.

This technique is based on a probabilistic model and employscooperativeprobabilitiesofcertaintermsandtheir respectivegroup for the estimation of theprobability of a certaingroup,withatextdocumentasinput.

In addition, the Support Vector Machine (SVM) is being presentedasaclassifiertohandlethechallengesofpattern

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 09 Issue: 10 | Oct 2022 www.irjet.net p-ISSN: 2395-0072

recognition between two groups. The Support Vector Machine(SVM)seekstodeterminetheoptimalhyperactive planemarginseparationbetweentwogroupsofdata.Itwas designedtosolveseparableinstances,butitcanbeexpanded tohandlelinearlynon-separableproblemsby transferring theoriginaldatavectortohigher-dimensionalspaces.Many researchers believe that the SVM classifier is the best methodtoutilisefortextclassificationandhaveusedit.

5. ANALYSIS STUDY:

Belowaresomeoftheprimaryusesforsentimentanalysis andopinionmining.

1) Identifying opinion spam: Customers may publish evaluationsregardingproductswithmaliciousintent.These reviewsmaybedividedinto"notspam"and"spam"material usingsentimentanalysisandopinionmining.

2) Buying a Product or Service: Using this method, consumersmayquicklyassessotherpeople'sopinionsand experienceswithanyProductorServiceandcomparerival brands.

3) Product or service quality improvement: Manufacturers may use this to gather both positive and negativefeedbackabouttheirgoods,whichhelpsthemraise thestandardoftheirgoods.

4) Marketing analysis: With their new government strategy, products or services be examined. These are all outcomes that may be attributed to group intelligent research.

5) Policy Making: Through sentiment analysis, policy makers may learn how citizens feel about a certain policy andusethisknowledgetodevelopnew,betterpoliciesthat aremoregearedtowardtheneedsofcitizens.

6) Decision-Making: Individuals' perspectives and backgroundsareacrucialcomponentinthedecision-making process. It offers assessed people’s opinion that may be properlyemployedformakingdecisions

6. CONCLUSION

Inconclusion,usingsentimentanalysisoropinionminingto mineavarietyofunstructureddatahasemergedasacrucial research issue. Sentiment analysis helps to create better goods,services,andeffectivebusinessmanagement.Inthe reviewarticle,relatedworkinthefieldofsentimentanalysis fromtheperiodwasgiven.

Futurestudyisrequiredtofurtherimprovetheperformance measurements. For any new applications that follow the principles of data mining, sentiment analysis or opinion miningcanbeused.

Even while the algorithms and techniques used for sentiment analysis are improving quickly and producing high-qualityfindings,manyissuesinthisareaofstudyare stillopen,anditmightbechallengingtospotfalsereviews justbyreadingthem.Sometimesfakereviewsaremistaken forgenuineonesandarealteredsothatnoonecantellwhat their true intentions were. Therefore, the identification of falsereviewsisanothercrucialareathatcallsfordeepdata miningapproaches.

REFERENCES:

[1] E. A. Stepanov and G. Riccardi, "Detecting General Opinions from Customer Surveys," 2011 IEEE 11th InternationalConferenceonDataMiningWorkshops,2011, pp.115-122,doi:10.1109/ICDMW.2011.63.

[2] Pankaj, P. Pandey, Muskan and N. Soni, "Sentiment Analysis on Customer Feedback Data: Amazon Product Reviews," 2019 International Conference on Machine Learning, Big Data, Cloud and Parallel Computing (COMITCon), 2019, pp. 320-322, doi: 10.1109/COMITCon.2019.8862258.

[3] P.Kherwa,A.Sachdeva,D.Mahajan,N.PandeandP.K. Singh,"Anapproachtowardscomprehensivesentimentaldata analysis and opinion mining," 2014 IEEE International AdvanceComputingConference(IACC),2014,pp.606-612, doi:10.1109/IadCC.2014.6779394.

[4] Rajkumar S Jagdale, Vishal S. Shirsath, Sachin Deshmukh,“SentimentAnalysisonProductReviewsUsing Machine Learning Techniques: Proceeding of CISC 2017”, Cognitive Informatics and Soft Computing, Advances in Intelligent Systems and Computing 768,https://doi.org/10.1007/978-981-13-0617-4_61.

[5] SatuluriVanaja,MeenaBelwal,Aspect-LevelSentiment AnalysisonE-CommerceData,2018.

[6] Z.Zhang,H.LiandW.Yu,"Fine-grainedopinionmining: An application of online review analysis in the express industry," 2017 3rd IEEE International Conference on ComputerandCommunications(ICCC),2017,pp.1498-1503, doi:10.1109/CompComm.2017.8322790.

[7] Barkha Bansal, Sangeet Srivastava, “Sentiment classificationofonlineconsumerreviewsusingwordvector representations”, Procedia Computer Science, Volume 132, 2018

[8] Chonghui Guo, Zhonglian Du, Xinyue Kou, “Mining Online Customer Reviews for Products Aspect-Based Ranking”, International Symposium on Knowledge and SystemsSciences,October2017,DOI:10.1007/978-981-106989-5_13.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 09 Issue: 10 | Oct 2022 www.irjet.net p-ISSN: 2395-0072

[9] Sasikala p, Mary Immaculate, Sheela Lourdusamy, "Sentimental Visualization: Semantic Analysis of Online Product Reviews Using Python and Tableau," 2020 IEEE InternationalConferenceonBigData(BigData),2020,pp.13,doi:10.1109/BigData50022.2020.9391769.

[10] C. S. R. Priya and P. Deepalakshmi, "Study on Online ReviewBasedConsumerSentimentalAnalysisusingMachine Learning Approaches," 2022 IEEE World Conference on Applied Intelligence and Computing (AIC), 2022, pp. 610616,doi:10.1109/AIC55036.2022.9848932.

[11] P. K. Singh, A. Sachdeva, D. Mahajan, N. Pande and A. Sharma, "An approach towards feature specific opinion mining and sentimental analysis across e-commerce websites," 2014 5thInternational Conference - Confluence the Next Generation Information Technology Summit (Confluence), 2014, pp. 329-335, doi: 10.1109/CONFLUENCE.2014.6949312.

2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal