Decision Support Portal for Access Certification

Abstract

Recommendation system is a technology which makes fast and intelligent business decisions. The goal of this paper is to present a recommender system that makes such decisions about user certifications in an organizational environment. It aims at analyzing the user database of a company in order to recommend how many privileges should be removed or assigned to the user based on the existing privileges as well as actions taken on them. Two types of recommendations have been explored, namely Static recommendations and interactive recommendations. The major difference between them is that static recommendation works on the entire existing data of an organization, users and their existing entitlements. It analyses that data and gives recommendations if any user has excessive or missing entitlements compared to his peers. On the other hand, interactive recommendation gives recommendation for dynamic data. When any action such as addition or removal is taken on an existing entitlement, the system analyzes a user’s current assignment and recommends addition or removal of entitlements similar to the ones on which action was taken. Outlier detection techniques and Associative rule mining are used respectively in order to build the system and tackle the problem of misuse of privileges in an enterprise setting.