Social Media Mining for Assessing Brand Popularity

Eman S Al-Sheikh, Mozaherul Hoque Abul Hasanat
2018 International Journal of Data Warehousing and Mining  
Businessesseektoanalysetheircustomerfeedbacktocomparetheirbrand'spopularitywiththe popularityofcompetingbrands.Theincreasinguseofsocialmediainrecentyearsisproducinglarge amountsoftextualcontent,whichhasbecomerichsourceofdataforbrandpopularityanalysis.In thisarticle,anovelhybridapproachofclassificationandlexiconbasedmethodsisproposedtoassess brandpopularitybasedonthesentimentsexpressedinsocialmediaposts.Twodifferentclassification
more » ... fferentbrands of3cosmeticproducts.Inaddition,sentimentquantificationhavebeenperformedusingalexiconbasedapproach.Basedontheoverallcomparisonoftheproposedmodels,theSVMclassifierhas thehighestperformancewith78.85%accuracyand94.60%AUC,comparedto73.57%and63.63% accuracy,80.63%and69.38%AUCoftheNBclassifierandthesentimentquantificationapproach respectively.Specificindicesbasedonclassificationandlexiconapproachesareproposedtoassess thebrandpopularity.
doi:10.4018/ijdwm.2018010103 fatcat:422heyty3ba4da2c5426ibxsce