Autonomic Knowledge Representation Construction for Software Protection Systems

Applications that are deployed using the web or that are based on web services are becoming more common and sophisticated, handling confidential and sensitive information, such as health records, credit card data, and social security numbers. Unfortunately, the wide accessibility of these applications, the critical nature of the data they handle, and the frequent lack of effective security mechanisms in their implementation have made these applications and services the favorite targets of attackers, who are looking for monetary gain or intelligence. We propose to develop an autonomic protection system based on machine learning techniques that builds a knowledge base associated with web applications and services and leverages this information to protect the modeled applications/services from novel attacks. In addition, the proposed system implements a model-driven request routing system that allows for the dispatching of possibly-malicious requests to versions of the web application/service that contain sanitized version of sensitive information. The resulting system will be able to automatically reconfigure the processing of requests so that both the impact of an attack and the consequences of a false positive are minimized.