| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 4.57 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Nos dias de hoje, o E-commerce tem tido uma crescente adesĆ£o tanto por parte das empresas como dos utilizadores, o que leva consequentemente a um aumento significativo no interesse por sistemas de recomendaĆ§Ć£o. Devido ao facto de as empresas conseguirem obter cada vez mais informaƧƵes, Ć© necessĆ”rio a utilizaĆ§Ć£o de um sistema capaz de processar as mesmas. Esses sistemas de recomendaĆ§Ć£o fazem assim uso de toda a informaĆ§Ć£o com a finalidade de recomendarem os produtos e serviƧos aos utilizadores. Ć na ideia do desenvolvimento de um sistema de recomendaĆ§Ć£o que esta dissertaĆ§Ć£o se assenta. SerĆ£o descritas tĆ©cnicas e mĆ©todos capazes de processar a informaĆ§Ć£o, em conjunto com as suas vantagens e limitaƧƵes. SerĆ£o tambĆ©m analisadas vĆ”rias mĆ©tricas capazes de avaliar as recomendaƧƵes dadas por um determinado sistema. SerĆ£o realizadas vĆ”rias avaliaƧƵes a diferentes mĆ©todos de recomendaĆ§Ć£o utilizando, diferentes mĆ©tricas como avaliaĆ§Ć£o, tendo como base vĆ”rios datasets de treino, com diferentes caracterĆsticas. Com todas essas avaliaƧƵes e anĆ”lises dos diferentes mĆ©todos, esta dissertaĆ§Ć£o culminou na implementaĆ§Ć£o de um sistema de recomendaĆ§Ć£o, sobre a forma de Web Service, que utiliza um mecanismo de recomendaƧƵes baseado em regras associativas e nos produtos mais populares. Com esta soluĆ§Ć£o Ć© possĆvel utilizar de uma forma otimizada a informaĆ§Ć£o obtidas acerca dos utilizadores tornando, simultaneamente, este um processo automĆ”tico para os clientes.
On the present day, E-commerce has an increasing adhesion both by companies and users, which leads to a significant increase in interest on recommendation systems. Since companies can obtain more and more information, it is necessary to use a system that is capable of processing that information. These recommendation systems thus make use of all information for the purpose of recommending products and services to users. Is on the idea of developing a recommendation system that this dissertation is based on. Techniques and methods capable of processing information will be described, alongside with its advantages and limitations. Various metrics capable of evaluating the recommendations given by the system will also be analysed. Several evaluations of recommendations methods will be carried out using different metrics, based on several training datasets, with different characteristics. With all these evaluations and analysis of the different methods, this dissertation culminated on the implementation of a recommendation system, as a Web Service, which uses a recommendation mechanism based on associative rules and the most popular products. With this solution, it is possible to use the collected information about users in an optimized way, making this process an automatic process for customers.
On the present day, E-commerce has an increasing adhesion both by companies and users, which leads to a significant increase in interest on recommendation systems. Since companies can obtain more and more information, it is necessary to use a system that is capable of processing that information. These recommendation systems thus make use of all information for the purpose of recommending products and services to users. Is on the idea of developing a recommendation system that this dissertation is based on. Techniques and methods capable of processing information will be described, alongside with its advantages and limitations. Various metrics capable of evaluating the recommendations given by the system will also be analysed. Several evaluations of recommendations methods will be carried out using different metrics, based on several training datasets, with different characteristics. With all these evaluations and analysis of the different methods, this dissertation culminated on the implementation of a recommendation system, as a Web Service, which uses a recommendation mechanism based on associative rules and the most popular products. With this solution, it is possible to use the collected information about users in an optimized way, making this process an automatic process for customers.
Description
Keywords
Sistema de recomendaĆ§Ć£o E-commerce Web Service