ANGEL: A HYBRID CONTENT-BASED FILTERING TOOL FOR PROTECTING TEENS’ SAFETY IN ONLINE SOCIAL NETWORK

Open Access
Author:
Zhang, Lei
Graduate Program:
Computer Science and Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
December 02, 2008
Committee Members:
  • Sencun Zhu, Thesis Advisor
  • Daniel Knifer, Thesis Advisor
Keywords:
  • Teens 
  • Social network
  • Filtering
  • Machine Learning
Abstract:
Since its emergence in 1997 [1] , online social networks have experienced dramatic evolution and played an increasingly important role in our life. With the enlarging population of members of various social networks, the functionality of themselves also enhanced so as to continuously boosting its further development. Participants in such online communities can not only find new friends of common interests by browsing their profiles , they can also stay connected with buddies via various asynchronized so-called virtual channels, like messaging, blogs comments, or by built-in applications.[2] While the social network (SN) owners are encouraging more and more people in getting their memberships and staying with them, there is also increasing concerns of security issues accompanying this new wave of internet blossom. Spamming problem is a long lasting issue existing in many field of web applications, conceivably, SN is one of them without exception [3]; Phishing attacks usually come with the spamming activity which lead even worse consequence once the victim has been deceived; other inappropriate materials like offensive contents [4], web-bully [5] and web scams [6] are also frequently observed by social network users. Security concerns mentioned above undoubtedly impair the user experience of web users, and are especially harmful to teens that usually have less capability in figuring out the malicious intensions behind the friendly disguise. Falsely trusted online friends could easily get valuable information from teens about themselves or even their family. This work addresses the possible threats on internet, evaluates state-of-art countermeasures, proposes and evaluates a light weight browser-based tool, named ANGEL, to help relieve the problem via hybrid-based content filtering. It is novel in the sense that, we find there is no similar approach discussed in previous literature or tools that could provide all around protection for teens’ safety in online social network.