Proposta de mecanismos para o problema combinatório de compras em lote de diferentes e-commerces: um estudo de caso do desenvolvimento de shopbots para Magic: The Gathering

Abstract

Shopbots services are common in Internet to compare prices from distinct providers. However, few of them find an optimal purchase configuration to a product list informed by users. In the card game Magic: The Gathering, there are many card retailers and users interested in acquiring a specific set of cards, so a tool to fetch purchase configurations with minimal cost would be interesting. In this work, it is proposed and evaluated mechanisms to perform price comparison for a product list in an e-commerces group, and to find purchase configuration with minimal cost. A case study was conducted with the Magic: The Gathering to achieve proposed goals. Product prices were collected and stored in a database, and they were used by cache mechanisms to mitigate response time from application. Three integer linear programing solutions were designed and implemented to get purchase configuration with minimal cost. The results demonstrated that developed techniques were able to find optimal solutions with response time guarantees. Empirical tests in the case study with 100 distinct products and 118 providers converged in about 25 seconds.