Detecting hazardous events from online news and social media
In this modern day, social media has seen immense growth that is constantly developing like nothing ever seen before. It has become the main source of information and entertainment for most of the population that owns a smartphone or a smart gadget. With that said, users all around the world has ena...
Saved in:
Main Author: | |
---|---|
Other Authors: | |
Format: | Final Year Project |
Language: | English |
Published: |
Nanyang Technological University
2024
|
Subjects: | |
Online Access: | https://hdl.handle.net/10356/177172 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
Summary: | In this modern day, social media has seen immense growth that is constantly developing like nothing ever seen before. It has become the main source of information and entertainment for most of the population that owns a smartphone or a smart gadget. With that said, users all around the world has enabled social media to become a significant source of real-time information and that includes reports of hazardous events such as fires, earthquakes and more. This information can be used to positively impact the world if handled correctly.
Through comprehensive research, systematic analysis and consistent experimentation, this project aims to develop a hazardous event-detecting system with the use of artificial intelligence, more specifically Natural Language Processing. Multiple corpora of text data from Twitter posts were obtained and used to develop and tune the hazard-detecting system to obtain high accuracy and relevant extraction of information from the classified tweets. With various pre-processing methods applied to the datasets, machine learning models were explored and developed to refine the overall performance of the detection model, a hazardous event detection system utilizing the Convolutional Neural Network, Density-Based Spatial Clustering Applications with Noise and Named Entity Recognition models was achieved. |
---|