gistfile1.txt

1. Set of questions or problems you hope your project will answer or address

This project addresses the problem of discovering and recommending unknown and unseen anomalies in access logs for large-scale security policy management. It presents a novel, non-standard recommender system called Helios that uses discrete categorical labels from access logs to build categorical combinations and offers a flexible and interpretable discovery engine for abnormal categorical combinations in access logs. 

2. Description of methodologies and approaches used in the project

The approach to be used would be the following three steps 
1. Constructing categorical combinations from discrete categorical labels in access logs
2. Using rank statistics based on the constructed categorical combinations to recommend highly abnormal patterns
3. Surface the reasoning behind the recommendation using visualization if possible (visualizing the vector space)

3. Expected results of the project

The expected results are to efficiently discover and recommend rules to block unknown and unseen anomalies in access logs for large-scale security policy management. 
The system is designed to offer a flexible and interpretable discovery engine for abnormal categorical combinations in access logs, which can be incorporated into existing security policy sets