Crowdview: uma plataforma crowdsourcing para gerenciamento temporal de entidades

Abstract

The web generates a lot of data about the same real-world object over time. In addition to this volume, the variety with which these data are presented grows substantially. In many cases this data is described in unstructured forms such as videos, images and texts. Algorithms for extracting data from unstructured forms are not yet accurate. In this case, it is appropriate to use the crowd to solve these tasks. In this context, crowdsourcing emerges as a paradigm shift in which the crowd, through open calls, provides solutions to specific problems. It can perform tasks distributed in different stages such as data collection, treatment, processing and analysis. These crowdsourcing systems produce a lot of data in a short amount of time. These initiatives also reduce time, operational costs and improve decision making. Given this scenario, the proposal in this doctoral thesis is the development of a crowdsourcing approach to extract and manage temporal characteristics of a real world object from unstructured information. This approach was implemented in a computer system called CrowdView. The case study is applied to the temporal management of urban forms. The analysis of the changes in characteristics in a chronological context can support decision making on the curation of urban forms within management of urban space.Também disponível on-line.