Re: 5.1.6 Rule language coverage <--: UCR Requirements Text

What about:

"To achieve widespread adoption, RIF dialects should cover shared features from 
many well-known rule languages"

-Chris


Paul Vincent wrote:
> Well, that's certainly *a* coverage requirement :)
> 
> But I think (i.e. my interpretation of) the meaning we are trying to
> convey is:
> 
>      Every standard RIF dialect should* support the rule processing
> semantics** and commonly used language attributes*** of the widely
> deployed rule engines that the dialect is meant to support. 
> 
> [[Explanation: 
> * = weaker requirement than "must", as this is difficult to measure
> ** = allows for RIF dialects that are not the focus of deployment at
> this time, or which are still considered R&D
> *** = this may be too onerous. 
> ]]
> 
> 	A rule engine's "processing semantics" is the functional
> algorithm used to interpret rules.
> 
> 	A rule engine's "commonly used language attributes" are the set
> of operators and functions**** that are used in some majority***** of
> rulesets that could be considered for interchange. 
> 
> 	A rule engine is considered "widely deployed" if it has over 100
> end-user deployments OR over 1,000 end-user developers. ****** 
> 
> [[**** = there may be some BLD-compliant term to use here.
> ***** = again, not measurable, but RIF will need to decide what is to be
> supported for this to be measurable
> ****** = open to debate on this definition]]
> 
> Paul Vincent
> TIBCO | Business Optimization | Business Rules & CEP
>  
> 
>> -----Original Message-----
>> From: Sandro Hawke [mailto:sandro@w3.org]
>> Sent: 05 June 2008 19:44
>> To: Paul Vincent
>> Cc: Christian de Sainte Marie; RIF WG
>> Subject: Re: 5.1.6 Rule language coverage <--: UCR Requirements Text
>>
>>
>> Let me try to paraphrase and slightly sharpen your proposed
> requirement:
>>     There must be at least one standard RIF dialect suited to
> conveying
>>     the rules used by each widely deployed rule engine.  (An engine is
>>     considered widely deployed if it currently has an installation and
>>     group of users at five or more separate organizations.)
>>
>> Does that get at what you're trying to say?   (Whether five is the
> right
>> number is kind of beside the point.)
>>
>> I don't think anything like this is practical.  For instance, I don't
>> expect any RIF dialect to be suited to conveying the rules used by
>> SWI-Prolog, which is certainly a widely deployed rule engine.  (I pick
>> it mostly because I know it the best.)
>>
>> So we could accept this requirement and then say we'll never meet it,
>> but I don't see the point in that.
>>
>> I would, however, advocate including text which explains why this is
>> *not* a requirement.
>>
>>     -- Sandro
>>
>>
>>>>> How about:
>>>>>
>>>>> RIF* must allow** rule interchange*** between common
> deployed****
>>> rule
>>>>> engines. =3D20
>>>>>
>>>>> * =3D3D RIF, the format, any extensions, and appropriate
> translators
>>>> =20
>>>> Standard extensions or third-party non-standard extensions?
>>> [PV>] Can a 3rd-party non-standard extension be part of / be
> regulated
>>> by a standard such as RIF? I'd assume extensions must be constrained
> to
>>> "standard extensions" (for what its worth).
>>>
>>> * =3D RIF, the format, any standard extensions, and appropriate
>>> translators
>>>
>>>> =20
>>>>> ** =3D3D subject to the development of appropriate compliant
>>> translators
>>>>> *** =3D3D interchange of rulesets against either a prespecified
> fact
>>> or =3D
>>>>> data
>>>>> model, or including said fact or data model
>>>>> **** =3D3D rule engines limited to individual research topics or
>>>>> institutions are assumed not to be both common and deployed;
> however
>>> RIF
>>>>> does not exclude these being covered.
>>>> =20
>>>> If standard extensions, then when do you think we can achieve
> this?
>>>> Certainly not in the next couple years, right?  We'd have to
> subsume
>>> the
>>>> prolog standardization work, etc.  And every time some rule vendor
>>> added
>>>> a feature, we would have failed in this goal until we caught up.
>>> [PV>] I'd say that these issues are inherent in RIF as a concept.
> The
>>> lack of metrics / difficulty in assessing whether this requirement
> is
>>> handled does not though detract from the general requirement for
>>> coverage. IMHO.
>>>
>>>> =20
>>>>      -- Sandro
>>>> =20
>>>>> Paul Vincent
>>>>> TIBCO | Business Optimization | Business Rules & CEP
>>>>> =3D20
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: public-rif-wg-request@w3.org
>>>>> [mailto:public-rif-wg-request@w3.org]
>>>>>> On Behalf Of Sandro Hawke
>>>>>> Sent: 03 June 2008 15:53
>>>>>> To: Christian de Sainte Marie
>>>>>> Cc: RIF WG
>>>>>> Subject: UCR Requirements Text
>>>>>> =3D20
>>>>> ...
>>>>>> =3D20
>>>>>> =3D20
>>>>>>> 5.1.6 Rule language coverage
>>>>>>>
>>>>>>> RIF must cover the set of languages identified in the
>>> _Rulesystem
>>>>>>> Arrangement Framework_. See the _Coverage_ section.
>>>>>> =3D20
>>>>>> Both those links are broken.  How about this:
>>>>>> =3D20
>>>>>>       RIF (with extensions) must cover all widely-deployed
> rule
>>>>>>       languages.
>>>>>> =3D20
>>>>>> =3D20
>>>>> ...
> 
> 

-- 
Dr. Christopher A. Welty                    IBM Watson Research Center
+1.914.784.7055                             19 Skyline Dr.
cawelty@gmail.com                           Hawthorne, NY 10532
http://www.research.ibm.com/people/w/welty

Received on Friday, 6 June 2008 14:03:43 UTC