The recent evolution of Information and Communication Technology (ICT) and mobile devices has strongly encouraged social participation as a tool for decision-support systems. These social participation tools are labelled as Participatory Geographic Information System (PGIS). The use of these tools has also extended to several domains – such as natural disasters, humanitarian crises, political conflicts – with the main aim to help affected populations and provide useful information for survival. Nonetheless, social participation tools present some drawbacks for managing non-structured information retrieved from large databases and Social Networks. The limitations concern either the need to understand knowledge in (almost) real time or data classification according to a specific domain. The present work aims at understanding the use of supervised classification models in situations of emergencies (i.e. disaster response) to classify message requests asking for/offering to help. To achieve the above aim we use machine learning techniques to compare classification models and evaluate their effectiveness and potentials to integrate them into existing PGIS systems. Main results suggest the existence of a relatively high accuracy of test and training classification by employing Random Forest, Neural Networks and Support Vector Machine (SVM) models. We argue in favour of supervised classification for its usefulness as a tool to be integrated in social participation for disaster response.

Integrating Supervised Classification in Social Participation Systems for Disaster Response. A Pilot Study

DE LUCIA, CATERINA
2017-01-01

Abstract

The recent evolution of Information and Communication Technology (ICT) and mobile devices has strongly encouraged social participation as a tool for decision-support systems. These social participation tools are labelled as Participatory Geographic Information System (PGIS). The use of these tools has also extended to several domains – such as natural disasters, humanitarian crises, political conflicts – with the main aim to help affected populations and provide useful information for survival. Nonetheless, social participation tools present some drawbacks for managing non-structured information retrieved from large databases and Social Networks. The limitations concern either the need to understand knowledge in (almost) real time or data classification according to a specific domain. The present work aims at understanding the use of supervised classification models in situations of emergencies (i.e. disaster response) to classify message requests asking for/offering to help. To achieve the above aim we use machine learning techniques to compare classification models and evaluate their effectiveness and potentials to integrate them into existing PGIS systems. Main results suggest the existence of a relatively high accuracy of test and training classification by employing Random Forest, Neural Networks and Support Vector Machine (SVM) models. We argue in favour of supervised classification for its usefulness as a tool to be integrated in social participation for disaster response.
2017
978-3-319-62401-3
978-3-319-62400-6
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11369/358857
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? 1
social impact