Finding Inadvertent Release of Information

The common interpretation of this topic is the accidental release of sensitive information such as e-mail or classified documents. However, this topic also addresses the issue that independently generated information, by multiple authors at different times, may together reveal more information than intended. The public website of a large organization may inadvertently disseminate sensitive information that is more than what is required using a need-to-know policy. The security of publicly disseminated information is a critical component of cyber-infrastructure and homeland security. The key problem in identifying such information involves inferences across multiple documents, since the information contained in a single document may on its own be innocuous. For example, one could synthesize the schedule of an important official by mining the web for announcements of his appearances. Techniques for detecting such vulnerabilities are complex and range from semiautomated methods using search engines to state-of-the-art research in text mining.