The DataRat was designed to help automate the capture, analysis, and reporting of the conversion processing rules needed to implement a transfer of data from a source data structure to a target data structure.

The DataRat stores two types of rules: navigation and transformation rules.  In DataRat, rationalization is a 2-step process: first, navigation rules are developed; second, transformation rules are developed.

Generally, the DataRat does not require the use of any specific “rules expression” language or syntax.  Nor does it employ any built-in functions or expressions.  Rules are composed as free form text.  Nevertheless, adopting some combination of relational algebra and procedural or logic flow constructs is highly recommended for rules expression.  Reports generated from the DataRat show the source objects that must be navigated in order to populate the target objects.  In addition, they show the rules and procedures for transforming source properties into target properties.

The first step, defining navigation rules, shows which source objects must be navigated to instantiate a valid target object record and to fully populate it.  In other words, it involves identifying the source objects needed to insert a valid instance into a target object and populate its non-key properties. As each source object is identified, navigation rules are attached.  This step involves extracting from one or more related source objects the data comprising the logical primary key attribute values of each target object in order to insure entity integrity.

The second step, defining property transformation rules, can begin once all source object navigation rules have been entered.  The DataRat does not enforce this practice, so analysts could actually enter a target object’s navigation rules and then proceed to map its property conversion rules, or map all property-to-property transformation rules first for all objects and then address all object navigation rules last.

To aid in the analysis process, transformation rules must be classified into one of the following six (mutually exclusive) conversion types:

1) New element  = No corresponding source element
2) 1:1 carryover = Direct domain value transfer
3) Physical field parse = Concatenated domains split into component domains
4) Aggregation = Component domains concatenated into one domain
5) Functional derivation = Parameter/method-based domain
6) Logical domain decomposition = “Hidden” information transformed into discrete domains