An Xlnet Based Target Dependent Sentiment Classification Along with a Smart Webapp for Post-disaster Management Using Django

In the aftermath of a calamity, a substantial number of individuals engage in communication through direct means or via social media platforms in order to seek assistance from governmental entities or organizations dedicated to disaster relief and recovery. In the event that an individual impacted by a particular issue chooses to communicate their thoughts or seek assistance through social media platforms or helpline services, there is a significant likelihood that their message may become obscured within the vast volume of incoming communications. The prevalence of excessive tweeting and the scarcity of individuals seeking assistance contribute to the challenge faced by organizations in manually filtering large volume of messages. So limiting their ability to respond promptly will be a challenging task. The current disaster management strategies are prevalent in various locations due to several drawbacks. To overcome the existing flaws, a smart web application is designed in which the peoples need will be met by meeting their requirements. It tracks the location where disaster occurs via tweets using target sentiment classification and the required items will be delivered as quick as possible which serves as a comprehensive platform that offers hotline services, contributing to the promotion of public welfare and the facilitation of efficient catastrophe management. This service serves as a platform for individuals to request essential items such as food, medicines, and groceries required at the time of disaster which helps in managing the welfare of the public efficiently during emergency situation. HTML, CSS, Bootstrap, Django, and Python has been used to develop the application. XLNet is used to perform classification of tweets. The maximum accuracy achieved though classification is 85% which is above all the other existing models.