| |
 |
|
[Semantic Web]Seven Golden Rules for a Web Rule Language |
Lee 发表于 2006/6/13 10:32:43 | GR 1: Relational databases are more important than FOL Many KR formalisms strictly (or blindly?) follow classical first-order logic and ignore the non-classical inference features and rule concepts, which have proven to be essential in relational databases, such as 3-valued connectives, nonmonotonic queries and (state-changing) trig-ger rules. A Web rule language cannot afford to ignore these fundamental information-processing concepts, which require abandoning classical logic.
GR 2: UML is more important than OWL The UML represents a larger body of information and knowledge modeling experience and expertise than OWL does. The UML includes an expressive language (OCL) for integrity constraints. These constraint expressions form also a kind of rule (‘integrity rules’) and should be covered by a Web rule language. Remarkably, the UML does also provide more support for advanced ontological constructs: e.g., it supports part-whole relationships (with aggregation and composition) and powertypes as classes whose instances are subclasses of another class (BiologicalSpecies and PassengerAircraft are examples of powertypes).
GR 3: Rules are not implications While an implication is an expression of a logical formula language, typically possessing a truth-value, a derivation rule does not possess a truth-value but has the role to generate derived sentences. There are logics, which do not have an implication connective, but which have a derivation rule concept. In standard logics (such as classical and intuit ionistic logic), there is a close relationship between a derivation rule (also called “sequent”) and the corresponding implicational formula: they both have the same models. For nonmonotonic rules (e.g. with negation-as-failure) this is no longer the case: the intended models of such a rule are, in general, not the same as the intended models of the corresponding implication.
GR 4: Web rules are not just Horn clauses This golden rule is a corollary of GR 3. Since Horn clauses are a very limited type of implication, and rules are not implications (according to GR 3), it follows that Web rules are not just Horn clauses. Web rules are rule expressions used in Web documents and in Web applications. They have to be much more expressive than Horn clauses (see also http://lists.w3.org/Archives/Public/www-rdf-rules/2001Sep/0079.html).
GR 5: Web rules should be able to express database rules In Web applications, we should expect similar uses of rules as in databases. This consideration suggests that a Web rule language has to accommodate . SQL assertions: integrity rules . SQL views: nonmonotonic derivation rules with 3-valued connectives and open and closed predicates . SQL triggers: reaction rules, which are limited to update events
GR 6: A Web rule language should allow to express and implement business rules Business rules refer to the hundreds, if not thousands, of policies, procedures and definitions that govern how a company operates and interacts with its customers and partners. Three basic types of business rules have been identified in the literature2: . Integrity rules: assertions that must hold in all evolving states and state transition histories of an enterprise viewed as a discrete dynamic system. Example: “The driver of a rental car must be at least 25 years old”. . Derivation rules: statements of knowledge that is derived from other knowledge by an inference or a mathematical calculation. Example: “A gold customer is a customer with more than $1MM on deposit”. . Reaction rules: expressions of policies specifying actions in response to events. Example: “When a share price drops by more than 5%, and the investment is exempt from profit tax, then sell it”.
GR 7: A Web rule language should allow multiple purposes, multiple languages and multiple semantics The Web is a pluralistic world , no matter if it is semantic or not. There will be multiple purposes, multiple languages and multiple semantics for Web rules. The real challenge is to develop an integrated metamodel, or abstract syntax, which supports this plurality.
|
|
|
| |
 | |
|
| Blog 信 息 |
blog名称:风落沙 日志总数:348 评论数量:550 留言数量:52 访问次数:1619544 建立时间:2005年1月28日 |
|
| 友 情 连 接 |
|

|
|
|

| |
|