Nowadays technology has appeared to be a very powerful ally in the fight against extreme natural phenomena. In this context, with social media establishing as the present-day technology of communication, platforms have been developed for an effective management of disasters and people and organizations seem to respond more actively to this new era. Clearly, all the above also emphasize the need of innovative approaches and new systems for a more effective disaster management. This dissertation deals with the exploration of the techniques and approaches based on user-generated data from social media that are currently in use for identification and management of natural disasters. It focuses on wildfires and elaborates on the factors that contribute to their expansion, providing at the same time an overview of the state of the art approaches for risk assessment and identification of such phenomena. Moreover, it is proposed a methodology and it was developed a system for collecting, filtering and geolocating Tweets from users. Through an analysis of the spatiotemporal aspect of the data, the approach aims at an early identification of a wildfire event. The aforementioned methodology was applied on a case of wildfires that occurred on the 23rd of Julia in the region of Attica in Greece, known as ‘Attica Wildfires 2018’. After the analysis of the collected dataset conclusions were reached and explained, challenges and difficulties were assessed and several maps were created for a detailed visualization of the results.
Ogni giorno la tecnologia si è dimostrata un alleato molto potente nella lotta contro fenomeni naturali estremi. In questo contesto, con i social media che si sono affermati come principale strumento di comunicazione, sono state sviluppate piattaforme per una gestione efficace dei disastri e le persone e le organizzazioni sembrano reagire più attivamente a questa nuova era. Chiaramente, tutto questo sottolinea anche il bisogno di approcci innovativi e nuovi sistemi per una gestione più efficace delle catastrofi. Questa tesi tratta le tecniche e gli approcci per l'identificazione e la gestione dei disastri naturali, basandosi su un’ analisi dei dati degli utenti generati tramite i social media. La discussione si concentra sugli incendi ed approfondisce i fattori che contribuiscono alla loro espansione, fornendo allo stesso tempo una panoramica degli approcci allo stato dell'arte per la valutazione del rischio e l'identificazione di tali fenomeni. Inoltre, viene proposta una metodologia ed è stato sviluppato un sistema per raccogliere, filtrare e geolocalizzare i tweet dagli utenti. Analizzando l'aspetto spaziotemporale dei dati, l'approccio mira ad identificare anticipatamente gli incendi. Questa metodologia è stata applicata all'incendio che si è verificato il 23 luglio nella regione dell'Attica in Grecia, noto come "Attica Wildfires 2018". Dopo aver esaminato il set di dati raccolto, sono spiegate le conclusioni raggiunte, vengono valutate le sfide e le difficoltà e sono mostrate diverse mappe per una visualizzazione dettagliata dei risultati.
Event identification via geolocation of user generated data on social media : the case of Attica wildfires
KARAMOSCHOS, ANTONIOS
2018/2019
Abstract
Nowadays technology has appeared to be a very powerful ally in the fight against extreme natural phenomena. In this context, with social media establishing as the present-day technology of communication, platforms have been developed for an effective management of disasters and people and organizations seem to respond more actively to this new era. Clearly, all the above also emphasize the need of innovative approaches and new systems for a more effective disaster management. This dissertation deals with the exploration of the techniques and approaches based on user-generated data from social media that are currently in use for identification and management of natural disasters. It focuses on wildfires and elaborates on the factors that contribute to their expansion, providing at the same time an overview of the state of the art approaches for risk assessment and identification of such phenomena. Moreover, it is proposed a methodology and it was developed a system for collecting, filtering and geolocating Tweets from users. Through an analysis of the spatiotemporal aspect of the data, the approach aims at an early identification of a wildfire event. The aforementioned methodology was applied on a case of wildfires that occurred on the 23rd of Julia in the region of Attica in Greece, known as ‘Attica Wildfires 2018’. After the analysis of the collected dataset conclusions were reached and explained, challenges and difficulties were assessed and several maps were created for a detailed visualization of the results.| File | Dimensione | Formato | |
|---|---|---|---|
|
Thesis_Karamoschos_Antonios_875051.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Thesis_text_Karamoschos
Dimensione
4.69 MB
Formato
Adobe PDF
|
4.69 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/148877