Minería de logs de una aplicación multiusuarío en línea

Abstract

This project proposes to perform data mining on activity logs of an application that is linked to the data of users in a social network. The application is developed in a virtual environment and adapts the concept of "In-Game Advertising", Advertising immersed in the game in order to have a higher profile of brands and interact directly with the customer. The processing is done in a distributed way using Hadoop as a map-reduce processing platform in conjunction with Pig and cloud computing services from Amazon. The end result of this process, is a report for each brand or product, which contains basic indicators for determining the acceptability of the brand or product within the context in which the game develops. The result analysis seeks to determine the best configuration to get the reports in the most efficient and the best time possible. The structure proposed presented acceptable results, although they can be improved.