Sumarizando informações na web através da identificação de tendências

Abstract

A research area that has been gaining space is the discovery of trending topics from data published on the Web. Some works perform the discovery process using relatively complex algorithms based on machine learning or using large volume of data gathered from internal databases. The purpose of this work is to show that it is possible to discover relevant topics through simple strategies and analyzing a small amount of information. To achieve this goal, we propose a tool called SE Trends (Search Engine Trends), which delegates the time consuming processing to a search engine. The tool applies information recovery techniques over pages returned from queries submitted to a search engine. Each term found is classified according to its importance, and the best classified terms are considered the most important topics. The experiments show scenarios where SE Trends returns terms that are indeed associated with issues of national concern.