[RuleML-all] RuleML 0.91 Naf Datalog and Integrity

Sat Mar 29 19:45:24 AST 2008

Hi David and All,

The turnstile |- is often used with either a set of formulas
(Rulebase |- Rulebase) or a 'query' (Rulebase |- Query) on the
right-hand side.

In RuleML 0.92 we should thus perhaps generalize

    Entails: ( oid?, (body | Rulebase), (head | Rulebase) )
    body-in-entails: (Rulebase)
    head-in-entails: (Rulebase)
to

    Entails: ( oid?, (body | Rulebase), (head | Rulebase | Query) )
    body-in-entails: (Rulebase)
    head-in-entails: (Rulebase | Query)

The Entails head can then really be this Query:

             <Query>
		    <Naf>
		      <And>
		        <Atom>
		          <Rel>gold</Rel>
		          <Var>object</Var>
		        </Atom>
		        <Atom>
		          <Rel>rusty</Rel>
		          <Var>object</Var>
		        </Atom>       
		      </And>
		    </Naf>
            </Query>

Please let us know if you'd have a problem with this change
in RuleML 0.92.

Other RuleML 0.92 changes are already in Reaction RuleML 0.2
(http://ibis.in.tum.de/research/ReactionRuleML/0.2).

Cheers,
Harold

---------------------------------------------------------------

PS: Maybe we shouldn't have used context-sensitive role names in
Entails, in the sense that, e.g., <head> has different content in
<Implies> and in <Entails>, i.e. roles are local to their classes
(http://www.ruleml.org/0.91/xsd/modules/connective_module.xsd):

. . .
<xs:group name="body.content">
	<xs:choice>
		<xs:element name="Atom" type="Atom.type"/>
		<xs:element name="And" type="And-inner.type"/>
		<xs:element name="Or" type="Or-inner.type"/>
	</xs:choice>
</xs:group>
<xs:group name="body-entails.content">
	<xs:choice>
		<xs:element ref="Rulebase"/>
	</xs:choice>
</xs:group>
. . .
<xs:group name="head.content">
	<xs:choice>
		<xs:element name="Atom" type="Atom.type"/>
	</xs:choice>
</xs:group>
<xs:group name="head-entails.content">
	<xs:choice>
		<xs:element ref="Rulebase"/>
	</xs:choice>
</xs:group>
. . .
<xs:group name="Entails.content">
	<xs:sequence>
		<xs:element ref="oid" minOccurs="0"/>
		<xs:choice>
			<xs:element name="body"
type="body-entails.type"/>
			<xs:group ref="body-entails.content"/>
		</xs:choice>
		<xs:choice>
			<xs:element name="head"
type="head-entails.type"/>
			<xs:group ref="head-entails.content"/>
		</xs:choice>
	</xs:sequence>
</xs:group>
. . .

These context-sensitive role names would go away in RuleML 0.92,
where we plan to change Implies' <body> to <if> and <head> to <then>,
so that we could keep, context-freely, Entails' <body> and <head>.

________________________________

From: ddossot at gmail.com [mailto:ddossot at gmail.com] On Behalf Of David
Dossot
Sent: March 29, 2008 6:57 PM
To: Boley, Harold
Cc: ruleml-all at ruleml.org
Subject: Re: [RuleML-all] RuleML 0.91 Naf Datalog and Integrity

Despite what I said below, I still can not implement an integrity
constraint in Naf Datalog 0.91. When I thought using an explicit body
would help me, I in fact misunderstood the Entails document model.

    Entails: ( oid?, (body | Rulebase), (head | Rulebase) )

What limits me is what I can use in a Rulebase (or head) to express the
integrity query. head is not enough, as it can only contain an Atom and
Rulebase is also not fitted as its content model:

    Rulebase: ( oid?, ( formula | Atom | Implies | Equivalent | Forall
)* )

does not allow Query. I could use an Implies, from which I would ignore
its head, but this is clunky.

Thanks for your help,
David