RE: [soa-rm-ra] RE: further thoughts on identifier and identity

Michael,

This is exactly in line with the discussion. I noted earlier the idea of a “sufficient set” of attributes and that the makeup of the set is context dependent. That gets into both what is the subject of identification and for what purpose. Strictly thinking that there is one all-purpose set of attributes for all occasions misses all of this.

Ken

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

Dr. Kenneth Laskey

MITRE Corporation, M/S H305 phone: 703-983-7934

7515 Colshire Drive fax: 703-983-1379

McLean VA 22102-7508

From: Mike Poulin [mailto:mpoulin@usa.com]
Sent: Wednesday, December 21, 2011 4:36 AM
To: Peter F Brown; Ken Laskey; Lublinsky, Boris; Bashioum, Christopher D
Cc: soa-rm-ra@lists.oasis-open.org
Subject: RE: [soa-rm-ra] RE: further thoughts on identifier and identity

If you do not mind the belate thought, I'd say that OO approach to identity (equality of all attributes) is rather artificial and only seems simple becuase it depends on complexity of each attribute that is not limited in any way.

In "applied mathematics", particularly, in the Cluster Analysis, only so-called "informative" attributes are counted. The informativness depends on 1) nature of clastering, 2) naure of challenged items. For example, in the US, SSN may be enough to define the identity of a person but in the UK NI number (job national insurance) is not enough becuase it does not cover all people living in the UK and even people working in the UK, i.e. additional attributes are need (speak from the personal experience of developing new governance pension scheme for the UK).

- Michael

----- Original Message -----
From: Peter F Brown
Sent: 12/21/11 06:05 AM
To: Ken Laskey, 'Lublinsky, Boris', 'Bashioum, Christopher D'
Subject: RE: [soa-rm-ra] RE: further thoughts on identifier and identity

I’ll take an action on this to draft such a paragraph…

Peter F Brown

Independent Consultant

www.peterfbrown.com

P.O. Box 49719, Los Angeles, CA 90049, USA

Tel: +1.310.694.2278

From: soa-rm-ra@lists.oasis-open.org [mailto:soa-rm-ra@lists.oasis-open.org] On Behalf Of Ken Laskey
Sent: Tuesday, 20 December, 2011 20:07
To: 'Lublinsky, Boris'; Peter F Brown; 'Bashioum, Christopher D'
Cc: soa-rm-ra@lists.oasis-open.org
Subject: [soa-rm-ra] RE: further thoughts on identifier and identity

yep! As I noted in my response to Peter, I think we have basic agreement and I think it is important to formulate that into a paragraph that conveys that agreement. I think identity is something fundamental when dealing with crossing boundaries and our audience needs this grounding.

Ken

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

Dr. Kenneth Laskey

MITRE Corporation, M/S H305              phone: 703-983-7934

7515 Colshire Drive                                    fax:        703-983-1379

McLean VA 22102-7508

From: Lublinsky, Boris [mailto:boris.lublinsky@navteq.com]
Sent: Tuesday, December 20, 2011 9:42 PM
To: Peter F Brown; Ken Laskey; 'Bashioum, Christopher D'
Cc: soa-rm-ra@lists.oasis-open.org
Subject: RE: further thoughts on identifier and identity

I am with Peter,

Everything has an identity, regardless whether we care to recognize it or not.

Identificator, on another hand is a handle for a given identity in a given context. A single thing always has an identity, but may have different identificators – in different contexts or systems

From: soa-rm-ra@lists.oasis-open.org [mailto:soa-rm-ra@lists.oasis-open.org] On Behalf Of Peter F Brown
Sent: Tuesday, December 20, 2011 6:42 PM
To: Ken Laskey; 'Bashioum, Christopher D'
Cc: soa-rm-ra@lists.oasis-open.org
Subject: [soa-rm-ra] RE: further thoughts on identifier and identity

Ken,

IS the identity of indiscernables a given? I hope so.

I think that you give too much weight to the role of identity in "determining what something is". We should not confuse identity with uniqueness. I would say that this is the function of identification and that identity merely provides a process towards providing an unambiguous handle to something unique.

I don't agree with you that:

"the identity of an entity is established when a sufficient set of attributes and associated values ... are all the same as .. a reference [set]"

- in context, that set of attributes might change: if there is only one green eyed person in a group, that is an attribute that is sufficient to establish uniqueness; if a second green-eyed person joins, you need to revise the set of attributes. The dictionary definition that you cite gets it right - it is about comparison, not absolute identification, which is the role of an identifier, not of identity.

Peter F Brown

Independent Consultant

www.peterfbrown.com

P.O. Box 49719, Los Angeles, CA 90049, USA

Tel: +1.310.694.2278

From: Ken Laskey [mailto:klaskey@mitre.org]
Sent: Wednesday, 16 November, 2011 20:43
To: Peter F Brown; 'Bashioum, Christopher D'
Cc: soa-rm-ra@lists.oasis-open.org
Subject: RE: further thoughts on identifier and identity

Thanks, Peter. This helped me refocus my thinking and incorporate some other suggestions. Try the following.

With the identity of indiscernibles () as a given, let’s look at what we seem to need and maybe, differently, what we actually need.

-          The typical use of “identity” is to determine who or what an entity is. I may want to make an unambiguous statement about the “identity” being a specific instance or it may be sufficient to identify a class to which the entity is a member.

-          The dictionary definition of identity (e.g. http://www.merriam-webster.com/dictionary/identity ) keeps referring to “same” or “sameness”, which implies a comparison. This is consistent with the class example in the previous point but the instance example has the typical feel of identity in isolation.

-          To establish the identity of an unknown (or unidentified) entity (or even the class of which the entity is a member), we need a set of attributes and associated values of a known entity (or known class) and then we compare those to the corresponding attributes and values of the unknown entity. If the values are the same for all compared attributes, then we can say we have identified the unknown entity as being the same as the known entity (or a member of the known class*).

* I’m assuming a class predicate could be a range and the corresponding instance predicate could be a value or a smaller range that is contained in the class range.

-          Mathematically, we conveniently say the comparison must hold for-all attributes but in practice we need to identify a sufficient set of attributes to be the subject of the comparison. So while height and weight are attributes for a person, not everyone with my height and weight is (the same entity as) me. An interesting example is what is required for a medical examiner to establish the identity of a dead body. What information is available and with what certainty can you establish identity?

-          This gets back to the connotation of identity as an absolute. What attributes and associated values do some authority need to say has satisfied a comparison so that an initially unknown entity has been identified as being a known one? I think this gives us our first point: The Identity of an entity is established when a sufficient set of attributes and associated values corresponding to the entity are all the same as an authorized set assembled and/or maintained as a reference.

-          This gives rise to the question of what is a “sufficient set of attributes”. It is likely defined by the same authority that is responsible for or has delegated the responsibility to assemble and/or maintain the reference set. There appear to be two classes of sufficient attribute sets, one which I will call “absolute” and the other which I will call “relative”. These follow from the original attempt at definitions.

-          A sufficient set of attributes is said to be absolute if the associated values remain the same independent of context or circumstance. The collection of attributes for identity is observable as those attributes not changing with context or circumstance.** For example, a person has a birth date; a server has a date when manufacturing/assembly is complete. The attributes making up identity vary by the class of the entity.***

** The identification of what is considered absolute may unintentionally be subject to context and may change over time.

*** Person is a class, server is a class …. There may be overlapping attributes among classes other than Person, but it is a distraction to try to define the class of Non-Person Entity (NPE).

-          A sufficient set of attributes is said to be relative if the associated values are dependent on context and/or circumstance. Relative attributes are defined and associated with one or more values within a context. For example, a person has a name used by some group (e.g. to my family in Philadelphia, I’m still (to the amusement of my children) Kenny), a server has a name assigned within a company’s cluster (e.g. each server has the name of a Muppet character).

-          The determination that a set of attributes is sufficient is itself based on context. For example, an undamaged driver license with a picture matching the likeness of the bearer and a birthdate before some designated date is sufficient in most (all?) states to establish identity of a member of the class who may be served an alcoholic beverage. The sufficient set of attributes may be a combination of absolute and relative attributes.

-          I had previously stated that Identity was based on a set of absolute attributes while Persona was based on a set of relative attributes or some combination of absolute and relative attributes. I would now consider a simpler formulation to be the bolded phrase above: The Identity of an entity is established when a sufficient set of attributes and associated values corresponding to the entity are all the same as an authorized set assembled and/or maintained as a reference.

-          Given this formulation, an entity may have many identities. An identity may be considered absolute if only absolute attributes make up the sufficient set used for comparison; an identity may be considered relative if some (or all) of the attributes making up the sufficient set are relative attributes.

-          Given this formulation, there is only identity and no need for the separate concept of Persona.

-          An identifier is an attribute that uniquely points to an entity within a context. Thus, an identifier is a relative attribute and is defined within a context, e.g. a URI or a GUID. An identity may have one or more unique identifiers; a unique identifier will point to only one entity within a context.

-          A Digital Identity is an Identity established within a digital environment. The attributes making up the sufficient set to establish Identity in this digital environment will all have a digital representation that can be used as part of automated comparison with the reference.

Ken

From: Peter F Brown [mailto:peter@peterfbrown.com]
Sent: Wednesday, November 16, 2011 7:34 AM
To: Ken Laskey; Bashioum, Christopher D
Cc: soa-rm-ra@lists.oasis-open.org
Subject: RE: further thoughts on identifier and identity

Ken,

Careful – “Some set of identifiers of a resource give rise to its identity” No, identity is an equality between two sets of properties, nothing more, nothing less. In many (in digital world, all) transactions, we accept to compare a small sub-set of properties that fall short of the “true”, “complete” identity of a person or entity in the real, analogue world. We need identity essentially to establish uniqueness and disambiguity, usually the latter: I don’t care if there are many Kens in the world, there is only one on the OASIS SOA-RM mailing list, so that is sufficient for a particular context.

We can say therefore that “some set of properties of an entity can be considered as validly identifying that entity in a particular context”. Those properties indeed serve to “identify” the entity but that is not the same as being the entity/resource’s identity. Something, btw, W3C still have all screwed up… ;-)

This is why I do not want to go into including Identity in any formal model because unless it states:

it will be wrong…. J

Peter

From: Ken Laskey [mailto:klaskey@mitre.org]
Sent: 03 November 2011 13:59
To: Bashioum, Christopher D; Peter F Brown
Subject: further thoughts on identifier and identity

Good discussion yesterday on this. let’s see if I can remember the insights that came to mind as I was going to bed last night.

-          Some set of identifiers of a resource give rise to its identity. More colloquially, they identify the resource.

-          Some identifiers are assigned, e.g. your parents gave you a name. Some identifiers are inherent, e.g. (we believe) your retinal scan is unique to your physical being.

-          The necessary and sufficient identifiers to uniquely identify a resource may vary depending on the situation. (Note, while the retinal scan of Mark Twain may have been unique to the physical being, it seems less relevant in identifying the historical figure. Also, while some situations would find the name Mark Twain to be a sufficient identifier, others might use Samuel Langhorne Clemens.)

-          For SOA, identifiers will be chosen/assigned that provide adequate identity; specifying those identifiers is beyond the scope of the SOA-RAF. (Chris: specifying those identifiers in our recent experience is an annoying diversion from settling on what we need to do with identity should it be established.)

-          Description will certainly reference identifiers that unambiguously WITHIN A CONTEXT point to what is being identified.

-          That said, I think our previous definitions of Identity, Identifier, and Description hold nicely.

Note, there are numerous places in the RAF where we still talk about identity. I think making these points in section 3 would clarify what we mean and what are the limits we consider in scope without drowning in philosophical arguments.

Ken

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

Dr. Kenneth Laskey

MITRE Corporation, M/S H305              phone: 703-983-7934

7515 Colshire Drive                                    fax:        703-983-1379

McLean VA 22102-7508

The information contained in this communication may be CONFIDENTIAL and is intended only for the use of the recipient(s) named above. If you are not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication, or any of its contents, is strictly prohibited. If you have received this communication in error, please notify the sender and delete/destroy the original message and any copy of it from your computer or paper files.

soa-rm-ra message