服务质量和顾客价值外文翻译文献.docx
《服务质量和顾客价值外文翻译文献.docx》由会员分享,可在线阅读,更多相关《服务质量和顾客价值外文翻译文献.docx(15页珍藏版)》请在冰豆网上搜索。
服务质量和顾客价值外文翻译文献
文献信息
文献标题:
AMethodforOpinionMiningofCoffeeServiceQualityandCustomerValuebyMiningTwitter(基于推文挖掘的咖啡服务质量和顾客价值的观点探勘)
文献作者及出处:
TakahashiS,SugiyamaA,KohdaY.Amethodforopinionminingofcoffeeservicequalityandcustomervaluebyminingtwitter[M]//Knowledge,InformationandCreativitySupportSystems.Springer,Cham,2016:
521-528.
字数统计:
英文1867单词,9106字符;中文3059汉字
外文文献
AMethodforOpinionMiningofCoffeeServiceQualityandCustomerValuebyMiningTwitter
AbstractInthiswork,wefocusoncustomervalueofJapanesecoffeeservicebyanalyzingthehugetweetdataset.Wesuggestthequantitativeevaluationmethodofcustomervaluebyusingthemorphologicalanalysisofthetweetsandgeneratedvaluedictionary.Oursuggestedmethodologycanbeappliedtoanyserviceandproducttoimprovethecommoditizationproblemfromtheviewpointofservicescience.
Keywords:
Customervalue;Naturallanguage;processingService;valueOpinionmining
1.Introduction
Recently,bigdatautilizationinsocialnetworkdata(e.g.Facebook,Twitter,Flicker)hasattractedmuchattentionfromresearchersandresearchdivisionsofcompaniesinordertoanalyzethemarketwithhigh“objectivity”andhigh“completeness”.Inthisstudy,wefocusontheTwitterdatafromtheviewpointofcustomersvalueofproductsandservicesinthecommoditizationofservice.Specifically,weexaminecoffeeserviceinJapanasanexampleofcommoditization.
Demandforcoffeebeanshasbeenincreasing,andcoffeeprovisioningmethodbecomingmorediversifyintheworldcoffeemarketsandservice.Inthedecadefromtheyear2000,coffeebeanconsumptionintheimportingcountrieshasincreasedtomorethan8%.Inaddition,coffeebeansconsumptionintheexportingcountrieshasincreasedby48%ormoreWiththeadventofaconveniencestorecoffee,acompetitioninthecoffeemar-kethasintensifiedinJapan.Thetotalnumberofsalesofconveniencestorecoffeehasincreasedto500millioncupsinlessthanoneyearfromthestartofsalesin2012.CoffeeconsumptioninJapanwas440thousandtonsin2013.Thisnumberincreasedby4.3%fromthepreviousyearbecauseofthehitofaconveniencestorecoffee.Therefore,commoditizationofSeattle-typecoffeeandconveniencestorescoffeehasprogressedintensifyotherresultofpricecompetition.Inaddition,anewtypeofcoffeeshopcalledThird-waveappearedaroundthewestcoastofNorthAmer-ica,andisspreadingallovertheworld.
ControllingcommoditizationincoffeeserviceisimportantforcoffeemarketnotonlyJapan,butalsowithintheemergingcountries.Inthisstudy,weexplorethepossibilitytocontrolthecommoditizationincoffeeservice.Therefore,weanalyzethevaluesofthecustomerswithinamongtheSeattle-typecoffeeandconveniencestorecoffee,toclarifytheneedsofthecustomersforcoffeeservice.
2.ProblemStatement
Thetypeofcoffeeserviceisclassifiedintothreecategories.Thefirstoneisaconvenient-store-stylecoffeethatisprovidedbyconveniencestores(CVS)andfastfoodshops(e.g.MaccafebyMcDonalds).Thiskindofcoffeehastheauthenticflavorandisinexpensiveprice.Thesecondtypeiscafeserviceprovidingrelaxingplace.ManynumberofSeattle-typecoffeepresentsuchkindofservice(e.g.TullysCoffee,StarbucksCoffee).Amongthese,byprovidingthethirdplacenoratworkorathome,thelargechainstoreStarbuckshavesucceededintosatisfytheneedsofcus-tomers.ThethirdtypeisaThird-wavecoffee.Third-wavecoffeeisonlyusessingleorigincoffeebeansthatwereharvestedfromseedlingsofasinglespecies,andismadebydrip.
InJapan,intensificationofpricecompetitivenesshasproceedingbetweenSeattle-typecoffeeandconveniencestorescoffee.Inthisstudy,weanalyzebythreepoints“taste”,“price”and“place”,thevaluesofthecustomerstotheCVSandStarbucks.UsingthetweetsofcustomersobtainedfromtheTwitterStreamingAPIdata,tobeanalyzedbythefollowingprocedure.
(i)ExtracttheJapanesetweetsthatarerelatedtocoffeeserviceusingtheTwitterStreamingAPI.
(ii)Generatedictionaryabout“taste”,“price”and“place”bytheJapanesetweetsextracted.
(iii)Mapafeaturespaceof“taste”,“price”and“place”aboutCVSandStarbucksusingthegenerateddictionary.
(iv)Fitofthestatisticalmethodstoanalyzetweetvectorsinthefeaturespace.
3.DataCollectionandFeatureExtraction
Inordertoextractthesenseofvaluesfromthetweetsofcustomers,wecreatedadictionaryaboutsenseofvalues(valuedictionary)relatedto“taste”,“price”and“place”.Inthedictionaryabout“taste”,“price”and“place”havebeenproduced,thetopfiveofthewordsareshowninTable1.
ExtractthetweetsaboutCVSandStarbucksamong118,855targettweets.Amongthetweets,wedefinedasCVSthatcontains[,convenience],andStarbucksthatcon-tains[,,Starbucks].Table2showsthecountsoftweetsrelatedtoCVSandStarbucksextracted.Thetargetdatais5,141tweetsoutof118,855Japanesetweetsaboutthecoffee.
WerepresentthetweetsofthetargetdatabyaVectorSpaceModelbasedonthevaluedictionary.AstheNwordspresentedinthevaluedictionarysuchasV1,V2,V3,…,VNtweetTisexpressedastweetvectortsuchasthefollowing.
Table1Thetop5frequentwordstodefinetheofdictionaryfromthetweetdatarelatedtocoffeetweet
Table2Countsoftweets
Class
Count
Starbucks
3612
CVS
1529
Total
5141
Inthesubsequentanalysis,weanalyzefromtweetvectors,whicharemappedtothefeaturespace.
4.Analysis
Ifcustomersarefocusedonvaluerelatedto“taste”,itcanbeexpectedthatthetweetvectorsincludevalueinthe“taste”dictionary.Wecomparevalueon“taste”ofcus-tomersaboutCVSandStarbucksthroughthetargettweets.Table3showsthenumberoftweetvectorscontainingthefeaturespacethatisgeneratedfromthedictionary.
WetestwithdifferenceofpopulationproportionusingthePearson’s𝜒2testsontheTable3a.FromtheresultsofthePearson’s𝜒2tests,therewasnodifferenceinthepopulationproportionatlevelofsignificance0.05.Becausethereisnodifferenceinthepopulationproportionoftweetrelatedtotaste,itcanbesaidthatconsumersarefocusedinthesamewayasCVSandStarbuckson“taste”.
Ontheotherhand,wefindthedifferenceinPearson’s𝜒2testson“price”and“place”atlevelofsignificance0.05showninTable3b,c.TheseresultsindicatethatCVScustomersarefocusedmorestronglyon“price”thanStarbuckscustomers.
Table3Thenumberoftweetsrelatedtothreecategories
F
T
Total
(a)Price
CVS
1280
249
1529
Starbucks
3017
595
3612
Total
4297
844
5141
(b)Place
CVS
1344
185
1529
Starbucks
3327
285
3612
Total
4671
470
5141
(c)Taste
CVS
1216
313
1529
Starbucks
2722
890
3612
Total
3938
1204
5141
Ontheotherhand,Starbuckscustomersarefocusedmorestronglyon“place”thanCVScustomers.
Weanalyzethatthecustomersarefocusedonwhatkindofvalueon“place”whentheybuyCVSorStarbucks.Table4showsfivewordswithagreatdifferenceatlevelofsignificance0.05inthepopulationproportion,whichcorrespondstothevaluedictionaryaboutplace.
Table4Differenceinthepopulationproportion
TherearemanypositivewordsforStarbuckscoffee,suchas“happy”,“high-quality”and“fashionable”.ItisevaluatedthattheplaceofferingStarbucksisahighqualitycomparedtotheCVS,andcustomersaresatisfied.Inaddition,manywordsaboutGenerationlike“adult”and“youth”haveappeared.
Aboveall,itisseenfromgenerationwordsappearedfromStarbuckstweetvectorsthatthirdplacehasbeenevaluatedfornotonlyequipmentbutalsogenerationsandpeoplewhogatherthere.Incontrast,theplaceofCVShasnotbeenevaluated,itisfoundthatcustomersarenotinterestedintheplacefordrinkingCVS.
Recently,positive-negativejudgementofSNStextbymachinelearningtechnicessuggestsomeusefulpredictionaboutmarketandpoliticaltrends.Takumuraetal.providetheJapanesesemanticorientationsofwordsbyconstractthelexicalnetwork.Intheirstudy,theytheemotionwastreatedasthespinofelectronsandsucceededinextractingsemanticorientationswithhighaccuracy,evenwhenonlyasmallnumberofseedwordsareavailable.Inourstudy,weestimatethesentimentalcoeffcientofcustumersbyJapanesesemanticorientationsoftweetwords.
SentimentalcoeffcientS(T)oftweetTiswrittenbyberow.
Here,scoreL(ai)andfrequencyF(ai)arecalculatedbyscoreofaverb,anounandanadjactivephrasea1,a2,a3,…,ai,…,atinsemanticorientationmap.L(ai)valueisfrom−1(negative)to1(positive).Distributionmapofsentimentalcoeffcientabout“price”and“place”indictionaryareshowninFig.1.
Fig.1Distributionmapofsentimentalcoefficient
Moreover,significantdifferencearenotobservedintheresultsoft-testaboutsentimentalcoefficientof“price”and“place”showninTable5.TheseresultsindicatethecastumervalueevaluatebynotonlysentimentalcoefficientofSNStextdata,butalsothequanitativeanalysisoftweetisimportantforconsidertheservicevalue.
Table5Sentimentalcoefficientof“price”and“place”
5.Discussion
Inthisstudy,wediscussthecustomersvalueofcoffeeservicebyanalyzingthehugetweetdataset.Wesuggestthequantitativeevaluationmethodofcustomer’svaluebyusingthemorphologicalanalysisofthetweetsandgeneratedvaluedictionary.
Thecharacterizesofallserviceandproductcanbeclassifiedintothreequalities:
searchqualities,experiencequalitiesandcredencequalities.Withcommoditization,thevalueofproductandservicefallintothesearchqualities.Originally,a