A study about mobility in Madrid based on Twitter datasets and R machine learning analytics

G. A. Valverde Castilla, V. López, B. González-Pérez

Social networks (SN) collect information about real time human activities. This information is huge and slanted. Some SN as Twitter generates a huge and heterogeneous data lake (Big Data ecosystem). Therefore, the knowledge about movements in the city is essential to learn about our society. In this paper we include all process to get information about routines Madrid citizens: create a system to get and store data from twitter usen a scheme of Retrieval Information over Big Data based on a hybrid structure, define a clustering system over geo-location, routines of some twitter users, and the application of some geographic visualization tools.

We propose a clustering algorithm implemented on Hadoop based on reinforcement learning that aims to determine dynamically real kind of activity place from a scoring in a distribution form defined by a Bayesian Network which takes nodes as the time slots of twitter publications and information from land registry.

Palabras clave: Clustering, Reinforcement learning, Bayesian Network, BigData, Social Network, Twitter API

X07.3 Big Data y Minería de Datos
7 de septiembre de 2016  15:40
Aula 21.08

L. Yuste Caro, R. González Martínez, D. Gómez González, C. Pareja Flores

R. Gonzalez Martinez, L. L. Yuste Caro, D. Gómez González, C. Pareja Flores

V. López, D. Urgelés, V. Hugo Mariscal, J. C. Anchiraico