Defesa de Tese de Doutorado de Maxwell Guimarães de Oliveira

postado em 28 de set. de 2016 05:51 por Coordenação da Pós-graduação em Computação da UFCG

Candidato(a): Maxwell Guimarães de Oliveira

Título do Trabalho: Ontology-driven Urban Issues Identification from Social Media


Cláudio de Souza Baptista

Cláudio Elizio Calazans Campelo

Data: 15/12/2016

Horário: 09:00:00

Local: Auditório do CEEI



Urban issues can be defined as problems related to the infrastructure on urban areas. Such issues are directly related to the urban space and may affect other people in or near to the area in question. For example, people can report a car parked on a footpath which is forcing pedestrians to walk on the road. Besides being related to the urban space, urban issues generally demand actions from city authorities. There are many Location-Based Social Networks (LBSN) in the smart cities domain worldwide where people complain about urban issues and local authorities are aware to fix the problems. Citizens became able to share valuable spatial, temporal and thematic information regarding urban neighborhoods. With the advent of social networks such as Facebook and Twitter, people tend to complain in an unstructured, sparse and unpredictable way, being difficult to let local authorities know about urban issues. Social media data, especially Twitter posts, photos, and check-ins, have played an important role in many fields, including city’s dynamics. However, discovering specific and relevant conversations about certain types of subjects is a challenge which gets worse on processing informal language, vernacular terms and all the noisy data found in social media streams. In this context, this research investigates computational methodologies in order to provide automated identification of urban issues shared in social media streams. Once identified, such issues can be semantically linked in a LBSN and become useful for citizens and authorities. Finally, this work proposes an ontology-driven approach which focuses on thematic and geographical aspects of urban issues. Experimental evaluation using crowdsourced data demonstrates the proposed approach is more feasible than traditional machine learning algorithms such as Support Vector Machines, Naive Bayes and Random Decision Forests.


Banca Examinadora: 

(Membros Internos)

Joseana Macêdo Fechine Régis de Araújo  

Leandro Balby Marinho

(Membros Externos)

Renato Fileto, Universidade Federal de Santa Catarina

Fabio Gomes de Andrade , IFPB