Using Deep Belief Network for Spammer Detection in Twitter Data

This project tries to detect spammers from twitter data using deep belief network. This work is inspired by Tzortzis, Grigorios, and Aristidis Likas. "Deep belief networks for spam filtering." Tools with Artificial Intelligence, 2007. ICTAI 2007. 19th IEEE International Conference on. Vol. 2. IEEE, 2007.

Tools used

MATLAB r2015a Deep Learning Toolbox !(https://github.com/rasmusbergpalm/DeepLearnToolbox)

Data used

The twitter data is collected by XiaHu for their paper titled "Hu, Xia, et al. "Social Spammer Detection in Microblogging." IJCAI. Vol. 13. 2013."

Files

• dbn_spammer_detect.m
• dbn_test_features.m
• dbn_train_features.m
• feature_extraction.m
• porterStemmer.m - stemming code (collected)
• rbm_train_features.mat - trained features
• vocab.mat - Vocabulary