Skip to content

Given dataset consists of tweets with textual messages and images, the task is to determine what kind of humanitarian information the tweets convey to help alert humanitarian organizations what kind of aid is required to save lives.

Notifications You must be signed in to change notification settings

qianxi5/Social-Media-Data-Classification-for-Disaster-Respnse

Repository files navigation

Social-Media-Data-Classification-for-Disaster-Respnse

During disasters, social media like Twitter is an important platform for people to post content to report updates about injured or dead people, infrastructure damage, and missing or found people among other information types. The information contained in Tweets helps humanitarian organization save lives and reduce damage. Therefore, we aim to classify the humanitarian information based on tweets with textual messages and images. We apply finetuned BERT model for text classification, ResNet-50 model for image classification, and decision level functions for multimodal classification. The classes of humanitarian information of tweets with texts and images are predicted well by our model.

About

Given dataset consists of tweets with textual messages and images, the task is to determine what kind of humanitarian information the tweets convey to help alert humanitarian organizations what kind of aid is required to save lives.

Topics

Resources

Stars

Watchers

Forks

Packages

No packages published