Autonomous Identity 2021.3.1


anomaly report

A report that identifies potential anomalous assignments.


A data source file type that includes application attributes.

as-is predictions

A process where confidence scores are assigned to the entitlements that users have.


A relationship between the user and an entitlement.


A data source file type that includes labels for the user and an entitlement.

association rule

The result of an Autonomous Identity machine learning process. It describes the rule for any given entitlement along with fequency, frequency union, and confidence score.


An action that an entitlement owner can do to approve a justification. Auto-certify indicates that anyone who has the justification is automatically approved for the entitlement.


An action that an entitlement owner can do to approve a justification. Auto-request indicates that anyone who matches these justification attributes but may not already have access should automatically get provisioned for this entitlement.

confidence score

A score from a scale from 0 to 100% that indicates the strength of correlation between an assigned entitlement and a user’s data profile. The score is derived from the ratio of frequency union over frequency.

data ingestion

A pre-analytics process that pushes the seven .csv files into the Cassandra database. This allows the entire training process to be performed from the database.

data sparsity

A reference to data that has null values. Autonomous Identity requires dense, high quality data with very few null values in the user attributes to get accurate analysis scores.

data validation

A pre-analytics process that tests the data to ensure that the content is correct and complete prior to the training process. This is run automatically.

driving factor

A single user attribute and its value that exceeds a confidence threshold level (e.g., 75%).


An entitlement is a unit of privilege, whether fine-grained or course-grained. A user or device with an entitlement gets access rights to specified resources.


A data source file type that includes the entitlement name, application-to-entitlement, roleowner, entitlement attributes.


The total occurrence of a rule within a user population.

frequency union

The total occurrence of a rule with a user population that has a specific entitlement.


A data source file type that includes job and department descriptions, HR name, and any identifying features of a user.

insight report

A report that provides metrics on the rules and predictions generated in the analytics run.


A process run after the as-is predictions that assigns confidence scores to all entitlements and recommends entitlements that users do not currently have. If the confidence score meets a threshold, set by the conf_thresh property in the configuration file, the entitlement will be recommended to the user in the UI console.


An external system, database, directory server, or other source of identity data to be managed and audited by an identity management system.


Representational State Transfer. A software architecture style for exposing resources, using the technologies and protocols of the World Wide Web. REST describes how distributed data objects, or resources, can be defined and addressed.


A collection of driving factors (i.e., justifications), where each driving factor is related to each other through an AND relationship.


A process that occurs after training that removes similar association rules that exist in a parent-child relationship. If the child meets three criteria, then it will be removed by the system. The criteria are: 1) the child must match the parent; 2) the child (e.g., [San Jose, Finance]) is a superset of the parent rule. (e.g., [Finance]); 3) the child and parent’s confidence scores are within a +/- range of each other. The range is set in the configuration file.


A multi-step process that generates the association rules with confidence scores for each entitlement. First, Autonomous Identity models the frequent itemsets that appear in the user attributes for each user. Next, Autonomous Identity merges the user attributes with the entitlements that were assigned to the user. It then applies association rules to model the sets of user attributes that result in an entitlement access and calculates confidence scores, based on their frequency of appearances in the dataset.

Copyright © 2010-2022 ForgeRock, all rights reserved.