- From: Dave Reynolds <der@hplb.hpl.hp.com>
- Date: Thu, 25 Aug 2005 14:14:32 +0100
- To: public-rule-workshop-discuss@w3.org
I would find it useful to get a better understanding of the intent of the WG, what scope of problems are intended to be solved and how does the technical proposal relate to that intent. At present the draft charter is aimed at a very wide scope but it's not clear to me that a single language or WG can really practically cover all of these. To be specific, it seems to me that there are at least three areas [*] which one might imagine a rules WG focusing on: 1. Vendor-neutral interchange of business rules to enable portability and reduce costs of integrating across multiple partners. 2. Higher level knowledge representation for the semantic web, extending beyond OWL. 3. Data integration, transformation and processing language for the semantic web. These have somewhat different constituencies and quite different technical requirements. The usage scenarios suggest a desire to cover all three (they arguably match up roughly in reverse order to the above list). (1) requires an interlingua and is explicitly the focus of the draft proposal but, as already discussed at length, FOL doesn't capture the semantics of existing business rules languages and transformations into it would be lossy. Other design choices would be affected if this were truly the focus, e.g. XML data would probably be much more important than RDF. [In this area, overlap with PRR seems more significant than acknowledged in the charter but that's a topic for a different forum.] (2) does not really require an interlingua since there aren't lots of such languages already in use. It would require strong OWL compatibility, and thus would presumably be undecideable. Ability to handle XML data directly would not be critical. This is the requirement that would seem to give the strongest motivation to a FOL-based approach. (3) requires something which is reasonably efficiently implementable, not a high level interlingua, works with RDF data (XML less critical) and at least layers with OWL[**]. Covering all three of these does not feel manageable to me, indeed any two may be sufficiently divergent to require an uncomfortably long-lived WG. A solution designed for one use may actually prove quite useful elsewhere but for a WG it seems to me one wants a quite narrow design centre. So if that framing makes any sense ... what's the *primary* intended focus of the WG? Dave [*] Clearly this is not a complete or unique partition of the space but my impression was that at least these requirements were represented at the workshop. [**] I'm referring to layered semantics in the sense of things like AL-log (if I'm understanding that correctly) rather than the side-by-side sense of building on top of DLP but that's a separate discussion for more knowledgeable folks.
Received on Thursday, 25 August 2005 13:14:51 UTC