An Abstract of the Thesis of Title: Adding Context and Value to Online Security Ratings by Combining Heuristics with Community Based Ratings. Adding Context and Value to Online Security Ratings by Combining Heuristics with Community Based Ratings

approved: ______________________________________________________ Carlos Jensen The Internet is growing rapidly in terms of websites, users and uses. People use the Internet for reference, shopping, social networking, communications, business and much more. Though the Internet is useful, there are many risks associated with its use, like malicious websites, identity theft, hateful content and fraudulent practices. Online safe surfing tools help Internet users stay safe from potential threats and have become common. These tools give security ratings for websites or automatically block and filter content. Many users rely on these ratings to identify dangers. This thesis examines different types of online safe surfing tools, analyzes their relative strengths and weaknesses, and ways to improve their quality. This thesis consists of two manuscripts. First, we conducted a study of the security ratings given by these tools. Based on these results, we identified two families of tools; 1) heuristic machine based and 2) community based tools. These differ both in analysis methodology and focus. Heuristics tools advise users about the technical foundations of a website but cannot be used to find whether the website engages in dubious practices, whether the content is for adults only, or whether the content is credible or reliable. Community based tools use user ratings, but determining the reliability of these is a problem. We proposed a method for extracting high quality information from user data that can be added with heuristic techniques. Second, we conducted a controlled lab experiment