| Titre : | Characterization of Flood Effects via Twitter | | Type de document : | projet fin études | | Auteurs : | Rania EL BOUHSSINI, Auteur | | Langues : | Français (fre) | | Catégories : | Ingénierie de web et Informatique mobile
| | Mots-clés : | Floods, natural human language, socioscope, machine learning algorithms, visualization. | | Index. décimale : | 1916/18 | | Résumé : | Social media has been highly implemented in our lives, about 8039 tweets per second [w1]. Business leaders and researches realized the importance of exploring this huge amount of data in many domains. During situational awareness, users tend to post messages in order to ask for help, prevent and report about the situational awareness. Analyzing these tweets could be of a tremendous help for local authorities to manage the disaster. Thus, the idea of considering citizens as sensors [0], which allows preventing excessive damage if the uploaded data is gently exploited. This thesis is mainly focused on floods being the most common type of disaster in the world [1], and most devastating. In all, floods have caused in France the death of more than a hundred peoples since the early 20th century [2] . Furthermore, flooding is also the first budget item of the French regime of natural disasters, representing 56% [3] of total costs of indemnifications provided. This fact motivates the Bureau des Recherches Géologiques et Minière to work on social sensing project applied to earthquakes, floods. My thesis concerns the flooding part. The purpose of this project is to build a model capable of classifying automatically the tweets according to a certain schema of classes, the output should be the information on the flooding effects on people, material infractures. In short, our goal is to use Twitter as a socioscope that connects local authorities to the affected zones without consuming much time and efforts. To do this, a vast and diverse data set need to uploaded, for all the French tweets during flooding time, a process of the human natural language should be done, and a comparison between the different machine learning algorithms is required. Finally, the build model should be evaluated and optimized, so we can present the outcome though some maps to visualize the work.
|
Characterization of Flood Effects via Twitter [projet fin études] / Rania EL BOUHSSINI, Auteur . - [s.d.]. Langues : Français ( fre) | Catégories : | Ingénierie de web et Informatique mobile
| | Mots-clés : | Floods, natural human language, socioscope, machine learning algorithms, visualization. | | Index. décimale : | 1916/18 | | Résumé : | Social media has been highly implemented in our lives, about 8039 tweets per second [w1]. Business leaders and researches realized the importance of exploring this huge amount of data in many domains. During situational awareness, users tend to post messages in order to ask for help, prevent and report about the situational awareness. Analyzing these tweets could be of a tremendous help for local authorities to manage the disaster. Thus, the idea of considering citizens as sensors [0], which allows preventing excessive damage if the uploaded data is gently exploited. This thesis is mainly focused on floods being the most common type of disaster in the world [1], and most devastating. In all, floods have caused in France the death of more than a hundred peoples since the early 20th century [2] . Furthermore, flooding is also the first budget item of the French regime of natural disasters, representing 56% [3] of total costs of indemnifications provided. This fact motivates the Bureau des Recherches Géologiques et Minière to work on social sensing project applied to earthquakes, floods. My thesis concerns the flooding part. The purpose of this project is to build a model capable of classifying automatically the tweets according to a certain schema of classes, the output should be the information on the flooding effects on people, material infractures. In short, our goal is to use Twitter as a socioscope that connects local authorities to the affected zones without consuming much time and efforts. To do this, a vast and diverse data set need to uploaded, for all the French tweets during flooding time, a process of the human natural language should be done, and a comparison between the different machine learning algorithms is required. Finally, the build model should be evaluated and optimized, so we can present the outcome though some maps to visualize the work.
|
|
Affiner la recherche Faire une suggestion
Characterization of Flood Effects via Twitter / Rania EL BOUHSSINI
