Internet users want their data on the Internet--for example, emails and Facebook data--to be well protected against malicious attacks. Current technologies either use end-to-end encryption for all data in the user’s Internet account, or use more advanced authentication techniques to protect access to the user account. These technologies are of great inconvenience to users. For example, each email should be decrypted before it is read or it is more difficult for valid users to access their accounts.

This paper presents a selective encryption technique that provides security to Internet users with acceptable convenience. The Cloudsweeper technique uses string searching or machine learning technology to automatically identify the security sensitive data in a user account, and then applies encryption to the identified security sensitive data only. The authors have developed a tool that can be publicly downloaded.

Cloudsweeper first searches through messages in a user’s email account for pieces of text that may be passwords, and then gives the account holder the option to encrypt each password while leaving the rest of the messages unchanged. Encrypted passwords are later decrypted when the emails are read. For identifying sensitive pieces of text, Cloudsweeper uses either handwritten regular expressions or technologies like Judo [1] to do more comprehensive term extraction.

If you are interested in Internet security, this paper is a good read, as you will get a clear understanding of how user account data is typically protected. It is also a good source for Internet security experts since it presents a novel perspective that helps broaden the horizon on providing security to Internet data. Service providers looking for a practical solution to guard customer data against malicious attacks should read the details on how to identify sensitive data and an evaluation of the Cloudsweeper technique [2]. According to the reference [2], Cloudsweeper identifies 96 percent of sensitive data.