Putrycz, Erik wrote:
> My current work focuses on extracting business rules from legacy source code. > From a legacy source code, we extract calculations and their context and, > to make the results understandable to business analysts, we link the > identifiers to any existing documents. Preliminary results have been > published at the RuleML 2007 and RuleML 2008 symposia. My interest in > RIF BLD is its use for exporting all the rules extracted from the > legacy code and transporting them to another tool for manipulating > and further transforming them.
This may be an unrealistic goal, as BLD only supports fairly simple rules, but depends obviously on the sophistication of the extraction.
Also, if you are extracting rules from programs, you may be more interested in PRD, the production rule dialect, which has an operational semantics. See http://www.w3.org/TR/rif-prd. In addition to having a more programming language style (i.e. operational) semantics, PRD is more in line with what is commonly referred to as "Business Rules".
> My comments on the RIF-BLD proposal are the following: > * Minor: the initials EBNF (Section 2.6) are never spelled out > * The RIF-BLD is sufficiently detailed and clear > * Rules are specified with the proper logic formalism, but I would like to > see more examples on the BLD/DTB interplay > * This specification offers a unique opportunity for exchanging rules > among tools and engines
Thanks for the feedback. We are developing more examples in the Test Cases framework, see an evolving list at http://www.w3.org/2005/rules/wiki/Category:Test_Case.
Hope this helps.
-The RIF WG