The CAS Data is a collection of FeatureStructures, where each FeatureStructure has an optional ID, a type (represented as a string), and a collection of features, which are attribute-value pairs. Feature names are strings, and their values may be primitives (String, integer, or float) or references, via ID, to one or more other FeatureStructures. Circular references are allowed.

Note on Arrays: Arrays are represented by ArrayFS, which is a subtype of FeatureStructure. Arrays are NOT primitive values. See FeatureStructure for more information.

FeatureStructures also have a property indexed, which determines whether the FeatureStructure should be added to the CAS's indexes if the CAS Data is converted to a CAS Object. The CasData itself does not provide indexes.