The Semantic Web and the Business Rules Approach ~ Differences and Consequences
My last column ended with a promise to discuss the differences in the formal specification languages used by the semantic web and business rules communities. I had observed that there are differences in the expression power of the semantic web rule languages and the business rules rule languages. The semantic web and business rule languages also differ in how they interpret particular expressions. I promised to give you more detail on these differences.
Keeping that promise turned out to be a great challenge. The longer I thought about it, the more I felt that the discussion might not belong in this column. Topics such as the following are appropriate for a textbook in formal logic
- closed world vs. open world assumption,
- higher order logic vs. first order logic,
- horn clause logic vs. predicate logic,
- deontic logic vs. non-modal languages,
- based purely on formal model theory vs. based on axioms,
- negation as failure vs. scoped negation as failure.
But are these useful topics for the average business rules professional?
I could try to explain to practitioners of the business rules approach what these differences are about, but that might not help a practitioner do his job. Are you really interested? I wonder... To give you a sense of this, let me illustrate with a short explanation of the first item in the list -- closed world vs. open world assumption.
Under a closed world assumption we assume that all relevant facts are known -- either directly or by derivation from known facts. If something is not known, we assume it to be false. This strategy is often applied to accepting an insurance policy, where rules govern when an insurance policy can be accepted. In a closed world, if we do not know (or cannot derive) that a given insurance policy is accepted, we assume that it is not accepted.
Under an open world assumption that conclusion is not valid. Instead, if we do not know (or cannot derive) that a policy is accepted the open world assumption considers is that it is 'unknown' whether or not the policy is accepted; we cannot conclude that it is 'not accepted'. Here is another example of the open world assumption that may be more intuitive. Consider the rule "if it rains the streets are wet." If we have no information about the weather conditions (raining or not) and ask "are the streets wet?" we cannot conclude 'not wet'. Instead, in an open world, when there are no facts to support a "yes, it's wet" answer all we can conclude is that that the wetness of the streets is not known. This is intuitive.
Rather than to continue on with each of the points above, what may interest you more are my observations regarding the consequences of these differences for those who work with business rules.
You do not need to understand these differences to work with business rules.
Business people have long been working with business rules without being aware of specific different logical interpretations that may exist for their business rule expressions. Did this lead to problems? Yes and no. Within a community, members generally agree on a common interpretation without conscious thought. However, it is good to be aware that problems in interpretation can arise when rules are communicated to another community, for example to an IT department.
You need to understand these differences to work with business rules.
The several standards for business rule specification that are in development differ on these points, and the person who will be choosing a language should know more about these differences to be able to make the best choice. Fortunately, that person does not need to be the businessperson who works with the rules.
Fundamentalism for one position undermines collaboration between the two communities.
I hope that the semantic web community and the business rules community will be working more closely together in the future. What is needed for that to happen is an attitude that we can listen and learn from each other. The new OMG standard SBVR is a result of such an attitude. SBVR includes carefully-selected language features and allows one to choose an interpretation that is best for a particular project.
 Silvie Spreeuwenberg, "The Semantic Web and the Business Rules Approach ~ Differences and Similarities," Business Rules Journal, Vol. 6, No. 7 (July 2005), URL: http://www.BRCommunity.com/a2005/b235.html
# # #