This document summarizes research on using data mining techniques to discover anomalies in firewall logs. It first provides background on firewalls and the issues around managing large and complex firewall rule sets. It then reviews different approaches that have been proposed for detecting anomalies like shadowing, correlation, generalization, redundancy, and irrelevance. Various data sources and rule generation methods are discussed. The document also summarizes several data mining techniques that have been applied to analyze firewall logs and detect anomalies, such as decision trees, association rule mining, and the Apriori algorithm. Finally, it concludes by stating that data mining provides a fast and effective way to gather useful information from large firewall logs to help optimize firewall rule sets.