| spelling |
Nivio ZivianiAdriano Alonso VelosoBerthier Ribeiro de Araujo NetoLeandro Balby MarinhoRicardo Baeza-yatesWagner Meira JuniorAnisio Mendes Lacerda2019-08-09T23:17:59Z2019-08-09T23:17:59Z2013-12-20http://hdl.handle.net/1843/ESBF-9GMN7JDaily-deals sites (DDSs), such as Groupon and Peixe Urbano, attract millions of customers in the hunt for offers at significantly reduced prices. The challenge of DDSs is to find the best match between deals and customers while generating as much revenue as possible. One important objective of a DDS is to improve the aggregated value customers give to emails, which should not be seen as spam. This thesis solves three different problems in order to guarantee revenue maximization and customer satisfaction. First, a method for predicting the number of coupons a deal is going to sell is proposed. Second, we present an email prioritization approach. Third, we introduce a new strategy for deals recommendation via email. All three methods improved the results of state-of-the-art algorithms for the tasks being addressed, with gains in precision varying from 7% to 21%, while reducing the number of emails sent in 40% without affecting the number of customers clicking the deals in emails.Daily-deals sites (DDSs), such as Groupon and Peixe Urbano, attract millions of customers in the hunt for offers at significantly reduced prices. The challenge of DDSs is to find the best match between deals and customers while generating as much revenue as possible. One important objective of a DDS is to improve the aggregated value customers give to emails, which should not be seen as spam. This thesis solves three different problems in order to guarantee revenue maximization and customer satisfaction. First, a method for predicting the number of coupons a deal is going to sell is proposed. Second, we present an email prioritization approach. Third, we introduce a new strategy for deals recommendation via email. All three methods improved the results of state-of-the-art algorithms for the tasks being addressed, with gains in precision varying from 7% to 21%, while reducing the number of emails sent in 40% without affecting the number of customers clicking the deals in emails.Universidade Federal de Minas GeraisUFMGSistema de recomendaçãoComputaçãoSistemas de recuperação de informaçãoRecommender SystemsDaily-deals sitesRevenue optimization and customer targeting in daily-deals sitesinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/doctoralThesisinfo:eu-repo/semantics/openAccessporreponame:Repositório Institucional da UFMGinstname:Universidade Federal de Minas Gerais (UFMG)instacron:UFMGORIGINALan_siomendeslacerda.pdfapplication/pdf1814368https://repositorio.ufmg.br/bitstream/1843/ESBF-9GMN7J/1/an_siomendeslacerda.pdf9533983e43695208734c1088d9962df1MD51TEXTan_siomendeslacerda.pdf.txtan_siomendeslacerda.pdf.txtExtracted texttext/plain220649https://repositorio.ufmg.br/bitstream/1843/ESBF-9GMN7J/2/an_siomendeslacerda.pdf.txtdb210264709a0298a108bc132b102fccMD521843/ESBF-9GMN7J2019-11-14 04:33:14.776oai:repositorio.ufmg.br:1843/ESBF-9GMN7JRepositório InstitucionalPUBhttps://repositorio.ufmg.br/oaiopendoar:2019-11-14T07:33:14Repositório Institucional da UFMG - Universidade Federal de Minas Gerais (UFMG)false
|