188宝金博页面版

Beau Crawford CS 5761 Final Project May 7, 2004 Gmail: A New Breed of Email Background Gmail is a newly offered web-based email system from Google. It can be thought of as an extension to its search engine service, which is commonly viewed as the most powerful search option available today. Gmail will operate in a manner similar to that of other web-based email services like Hotmail and Yahoo. Competing services like these typically offer users only a few megabytes of storage capacity. In complete contrast, Gmail encourages users to not “throw anything away” by offering them one gigabyte (1024 megabytes) of email. In having that amount of information stored, Google believes that users will require a non-traditional way of browsing their emails or, as they put it, a “search, don’t sort” approach. Instead of relying on multiple sub-directories to house archived email, Google will offer users the capabilities of its search engine technology to sift through their archived email. Problem Description As users use Google to explore their archived email, in a manner similar to web searching, the Gmail system will be analyzing both their search criteria and the content of the emails being searched. The returned list of results will then allow Google to generate and display relevant advertisements in a non-invasive manner. For example, searching for an email about an upcoming vacation with a friend is likely to cause ads for hotels and rental cars to be displayed. A similar sequence of events will take place as users read new email and browse old emails on their own. The difference here however will be that the email content will be “blindly” analyzed by the Gmail system – meaning that it doesn’t necessarily know what the user is interested in. The Gmail system will attempt to determine the key subjects of the email and then display relevant ads to the user. The focus of this project will be on the “blind” analysis of emails. My system will have the capability to analyze the body of an email and programmatically find its main topics. From there, it will attempt to link them to pre-built advertisements that directly relate. Henzinger, Chang, Milch, and Brin discuss a system that associates television content with web articles the listener might be interested in [1]. A variety of algorithms, including things like document frequency, are discussed in this article. My system deals with a similar concept except that, instead of TV content, it will be analyzing email and, instead of web articles, it will be displaying ads. My system will not directly implement any of these algorithms. Instead, I will be relying on their system as a conceptual foundation for my own ideas. Practical Applications Practical applications for such a system are somewhat limited. This is due to the fact that a system like Gmail requires you to be granted permission to electronically analyze private information. The system itself is basically identical to a Web-based advertisement linking