Support Efficient, Scalable, and Online Social Spam Detection in System

Authors

Hailu XuFollow

Document Type

Dissertation

Degree

Doctor of Philosophy (PhD)

Major/Program

Computer Science

First Advisor's Name

Liting Hu

First Advisor's Committee Title

Committee chair

Second Advisor's Name

S. S. Iyengar

Second Advisor's Committee Title

Committee member

Third Advisor's Name

Deng Pan

Third Advisor's Committee Title

Committee member

Fourth Advisor's Name

Alex Afanasyev

Fourth Advisor's Committee Title

Committee member

Fifth Advisor's Name

Gang Quan

Fifth Advisor's Committee Title

Committee member

Keywords

Social spam detection, scalable system, stream system, online social network

Date of Defense

6-24-2020

Abstract

The broad success of online social networks (OSNs) has created fertile soil for the emergence and fast spread of social spam. Fake news, malicious URL links, fraudulent advertisements, fake reviews, and biased propaganda are bringing serious consequences for both virtual social networks and human life in the real world. Effectively detecting social spam is a hot topic in both academia and industry. However, traditional social spam detection techniques are limited to centralized processing on top of one specific data source but ignore the social spam correlations of distributed data sources. Moreover, a few research efforts are conducting in integrating the stream system (e.g., Storm, Spark) with the large-scale social spam detection, but they typically ignore the specific details in managing and recovering interim states during the social stream data processing. We observed that social spammers who aim to advertise their products or post victim links are more frequently spreading malicious posts during a very short period of time. They are quite smart to adapt themselves to old models that were trained based on historical records. Therefore, these bring a question: how can we uncover and defend against these online spam activities in an online and scalable manner? In this dissertation, we present there systems that support scalable and online social spam detection from streaming social data: (1) the first part introduces

Oases, a scalable system that can support large-scale online social spam detection, (2) the second part introduces a system named SpamHunter, a novel system that supports efficient online scalable spam detection in social networks. The system gives novel insights in guaranteeing the efficiency of the modern stream applications by leveraging the spam correlations at scale, and (3) the third part refers to the state recovery during social spam detection, it introduces a customizable state recovery framework that provides fast and scalable state recovery mechanisms for protecting large distributed states in social spam detection applications.

Identifier

FIDC009021

Previously Published In

Xu, Hailu, Liting Hu, Pinchao Liu, and Boyuan Guan. "Exploiting the Spam Correlations in Scalable Online Social Spam Detection." In International Conference on Cloud Computing, pp. 146-160. Springer, Cham, 2019.

Xu, Hailu, Liting Hu, Pinchao Liu, Yao Xiao, Wentao Wang, Jai Dayal, Qingyang Wang, and Yuzhe Tang. "Oases: an online scalable spam detection system for social networks." In 2018 IEEE 11th International Conference on Cloud Computing (CLOUD), pp. 98-105. IEEE, 2018.

Xu, Hailu, Boyuan Guan, Pinchao Liu, William Escudero, and Liting Hu. "Harnessing the nature of spam in scalable online social spam detection." In 2018 IEEE International Conference on Big Data (Big Data), pp. 3733-3736. IEEE, 2018.

Recommended Citation

Xu, Hailu, "Support Efficient, Scalable, and Online Social Spam Detection in System" (2020). FIU Electronic Theses and Dissertations. 4499.
https://digitalcommons.fiu.edu/etd/4499