Re: The Object Primitive

Hi,

Accuracy is important.  If all data was accompanied by some measure of its accuracy, which in turn is a measure of the accuracy of the instruments, the mathematical model, even of alterations made in the data, we would probably not trust any numbers we see because the compound errors would be enormous.  Best to keep such things quiet, perhaps so we can live in blissful ignorance.

/Roy.
--
--------------------------------------------
Dr. Roy C. Davies, The VR Guy.
--------------------------------------------
Managing Director, LOOK-HERE IP Holdings Ltd.
Consultant and Managing Director, The Flexible Reality Studio Ltd.
Senior Research Fellow, Auckland University of Technology (AUT)

EMAIL: roy.c.davies@ieee.org, roy.c.davies@flexstudio.co.nz, roy.c.davies@aut.ac.nz, roycdavies@mac.com

LINKEDIN: http://nz.linkedin.com/in/roycdavies
MYSPACE: http://www.myspace.com/roycdavies
FACEBOOK: http://www.facebook.com/roy.c.davies
TWITTER: http://twitter.com/roycdavies

SKYPE: roycdavies
MSN: roy.c.davies@ieee.org
ICQ: 2557565
YOUTUBE: drroycdavies

PH: +64 (0)21 795294, +64 (0)9 8338360
WEB: www.flexstudio.co.nz, www.look-here.info, www.humanitycomputer.org, www.forwardthinking.org.nz, www.colab.org.nz

On 15/02/2011, at 10:59 PM, Thomas Wrobel wrote:

> Yes, this is somewhat of a concern of mine.
> The thing with AR is there is such a large number of things that will
> be useful to the end "result" for the user, and such a large number of
> factors that could influence even how that "result" is displayed we
> could get lost in an factual web of parameters,fields and metadata.
> Reading over the papers there was many times I was thinking that weeks
> could be spent discussing any one of the subjects people were bringing
> up. I think trustworthyness.....certainly not exclusive to AR....falls
> into this category.
> 
> I was thinking of accuracy earlier merely representing the accuracy of
> a specified location point according to the device that submitted
> it....but even that depends on the publishing source being
> "trustworthy", so even something as simple as that raises issues.
> 
> I can only suggest we have to deal with the fundamental things first
> as best way can (How are we linking things to reality? What formats
> are we linking with? etc), with a mind to new fields and expansion in
> the future.
> 
> -Thomas
> 
> 
> ~~~~~~
> Reviews of anything, by anyone;
> www.rateoholic.co.uk
> Please try out my new site and give feedback :)
> 
> 
> 
> On 15 February 2011 10:07, Jens de Smit <jens@layar.com> wrote:
>> Hey all,
>> 
>> Blatantly ignoring the "slippery slope ahead" sign...
>> 
>> Aren't these "trustworthiness" and "accuracy" properties applicable to
>> almost any information you find on the internet (or, for that matter,
>> in the real world)? Also, what I gather from the previous discussion
>> is that we require a service to verify this (subjective) claim of
>> accuracy and trustworthiness. Combine the above two statements and I
>> see an opportunity for a service that can ascertain accuracy and
>> trustworthiness claims for any resource, as long as that resource is
>> uniquely identifiable. I am _still_ woefully ignorant with respect to
>> these subjects, but this sounds very Internet of Things/Semantic Web
>> to me.
>> 
>> Do people agree, are there any known active efforts in this area and
>> do _we_ want to move further on this subject?
>> 
>> Regards,
>> 
>> Jens
>> 
>> On Mon, Feb 7, 2011 at 7:33 PM, Seiler, Karl <karl.seiler@navteq.com> wrote:
>>> Thomas,
>>> 
>>> All great insights. You are correct that "how" the trust is calculated or gamed is a potential problem. I was not so concerned with how trust was generated rather that is would be so nice to know if someone knew for sure the place was there, versus they only suspected the place was there, especially if it is a hospital emergency entrance, or an airport departure gate, or a court house, or ....
>>> 
>>> _______________________________
>>> Karl Seiler
>>> Director Location Technology & Services
>>> NAVTEQ - Chicago
>>> (T)  +312-894-7231
>>> (M) +312-375-5932
>>> www.navteq.com
>>> 
>>> 
>>> -----Original Message-----
>>> From: Thomas Wrobel [mailto:darkflame@gmail.com]
>>> Sent: Monday, February 07, 2011 10:56 AM
>>> To: Seiler, Karl
>>> Cc: cperey@perey.com; Jens de Smit; public-poiwg@w3.org
>>> Subject: Re: The Object Primitive
>>> 
>>> Ah, right.
>>> I get what your saying here now - there is a clear distinction between the two.
>>> 
>>> But surely that sort of trustworthy system would need an
>>> infrastructure of validation - with the trustworthiness field only
>>> settable by a...ahum...trustworthy party.
>>> 
>>> I mean, on a open network with anyone hosting and submitting content
>>> (as I believe we are all aiming for), surely anyone could just set
>>> their trustworthy on their POI to max by default :? Thus I'm not sure
>>> how useful it will be in real world situations.
>>> You could, of course, have wikipedia-like database's with multiple
>>> user validation. In this way the service owner becomes the manager of
>>> that field.  But I think they would be better of managing their own
>>> filtering/confirmation/"degree of certainty" systems. They might, for
>>> example, want to add information about the number of users confirming
>>> or rejecting a result - a thumb up/down system, a automatic popup, or
>>> perhaps even a api that lets people double check other placemarks
>>> against a common databases.
>>> Yet another option might be for the user's client itself to decide the
>>> trusthworthyness of the input based on the sources origin. (and
>>> different users might have different preferences to that)
>>> 
>>> There seems many different ways of doing it so I'm not sure how
>>> usefull building one method, say "trustworthyness percentage" into the
>>> POI (which is effectively an "output spec") is.
>>> 
>>> -Thomas
>>> 
>>> ps. Sorry if some of this is covered in papers people have submitted,
>>> I've only read about 5 so far. I'll try to get them all and have some
>>> feedback in time for conference.
>>> 
>>> ~~~~~~
>>> Reviews of anything, by anyone;
>>> www.rateoholic.co.uk
>>> Please try out my new site and give feedback :)
>>> 
>>> 
>>> 
>>> On 7 February 2011 17:20, Seiler, Karl <karl.seiler@navteq.com> wrote:
>>>> It is a tricky topic.
>>>> 
>>>> A GPS coordinate has a "high" degree of accuracy
>>>> An address has a "lower" degree
>>>> A center point of a general city center has lower accuracy still, for "where is this place."
>>>> 
>>>> If the GPS coordinate comes from a highly trusted corroborated source - its trustworthiness is high
>>>> If the source of the GPS coordinate was low-trust individual community input - its trustworthiness can be lower.
>>>> Both with the same relative degree of accuracy.
>>>> 
>>>> _______________________________
>>>> Karl Seiler
>>>> Director Location Technology & Services
>>>> NAVTEQ - Chicago
>>>> (T)  +312-894-7231
>>>> (M) +312-375-5932
>>>> www.navteq.com
>>>> 
>>>> 
>>>> -----Original Message-----
>>>> From: Thomas Wrobel [mailto:darkflame@gmail.com]
>>>> Sent: Monday, February 07, 2011 9:43 AM
>>>> To: Seiler, Karl
>>>> Cc: cperey@perey.com; Jens de Smit; public-poiwg@w3.org
>>>> Subject: Re: The Object Primitive
>>>> 
>>>> Would not trustworthyness be more like accuracy? Accuracy gives us
>>>> some distinctive units to use (ie, distance radius), Trustworthyness
>>>> implies a more estimated/subjective thing which I'm not sure will be
>>>> especially usefull for software.
>>>> 
>>>> In many case's a device placing a POI will know its own accuracy - if
>>>> a device is on gps it knows its margin of error based on the number of
>>>> satellites in view. In this way even the degree of error can be to
>>>> some extent "accurate".
>>>> So anyone createing content directly "on site" could have their
>>>> software automaticly add this field
>>>> 
>>>> -Thomas
>>>> 
>>>> ~~~~~~
>>>> Reviews of anything, by anyone;
>>>> www.rateoholic.co.uk
>>>> Please try out my new site and give feedback :)
>>>> 
>>>> 
>>>> 
>>>> On 7 February 2011 15:22, Seiler, Karl <karl.seiler@navteq.com> wrote:
>>>>> IDs.
>>>>> 
>>>>> 
>>>>> 
>>>>> I included (and you inherited) the idea of associated IDs. My thought here
>>>>> was not to represent relationships, only to support the concept that IDs for
>>>>> the same location can change over time or go in and out of use, and this
>>>>> attribute provides linkage to the old IDs.
>>>>> 
>>>>> 
>>>>> 
>>>>> Sometimes IDs are retired and replaced, or are depreciated and deleted.
>>>>> Since we are not going to provide any inherent sync-ing mechanism then one
>>>>> system can be carrying old / bad IDs. This attribute can convey a reference
>>>>> bad to old IDs, allowing out of sync systems to bridge their persistent IDs
>>>>> forward.
>>>>> 
>>>>> 
>>>>> 
>>>>> Admittedly kind of obscure.
>>>>> 
>>>>> 
>>>>> 
>>>>> Trustworthiness
>>>>> 
>>>>> Trustworthiness : degree of certainty the author has in the accuracy of the
>>>>> 
>>>>> object
>>>>> 
>>>>> 
>>>>> 
>>>>> What makes an object more or less accurate?
>>>>> 
>>>>> I think mobile things expect and require a higher degree of accuracy and
>>>>> therefore trustworthiness in the confidence in their location. The use cases
>>>>> get rare but render something like: we know it is moving, near or around
>>>>> this building, but we really do not know where it is right now.
>>>>> 
>>>>> 
>>>>> 
>>>>> FYI - I did post a Location Primitive revision.
>>>>> 
>>>>> 
>>>>> 
>>>>> _______________________________
>>>>> 
>>>>> Karl Seiler
>>>>> 
>>>>> Director Location Technology & Services
>>>>> 
>>>>> NAVTEQ - Chicago
>>>>> 
>>>>> (T)  +312-894-7231
>>>>> 
>>>>> (M) +312-375-5932
>>>>> 
>>>>> www.navteq.com
>>>>> 
>>>>> 
>>>>> 
>>>>> From: public-poiwg-request@w3.org [mailto:public-poiwg-request@w3.org] On
>>>>> Behalf Of Christine Perey
>>>>> Sent: Sunday, February 06, 2011 4:07 AM
>>>>> To: Jens de Smit
>>>>> Cc: public-poiwg@w3.org
>>>>> Subject: Re: The Object Primitive
>>>>> 
>>>>> 
>>>>> 
>>>>> Hi Jens, thanks for your excellent points!
>>>>> 
>>>>> my replies in-line.
>>>>> 
>>>>> On 2/2/11 1:53 PM, Jens de Smit wrote:
>>>>> 
>>>>> Hey,
>>>>> 
>>>>> 
>>>>> 
>>>>> Some feedback/thoughts/discussion points
>>>>> 
>>>>> 
>>>>> 
>>>>> On Sat, Jan 29, 2011 at 9:59 AM, Christine Perey <cperey@perey.com> wrote:
>>>>> 
>>>>> I took the locations primitive which Karl provided
>>>>> 
>>>>> 
>>>>> 
>>>>> http://lists.w3.org/Archives/Public/public-poiwg/2011Jan/0017.html
>>>>> 
>>>>> 
>>>>> 
>>>>> as well as the suggestions from Roy Davies
>>>>> 
>>>>> 
>>>>> 
>>>>> http://lists.w3.org/Archives/Public/public-poiwg/2011Jan/0007.html
>>>>> 
>>>>> 
>>>>> 
>>>>> and created, based on those, a laundry list of what could be included in the
>>>>> 
>>>>> "objects primitive".
>>>>> 
>>>>> 
>>>>> 
>>>>> As Roy suggests and we discussed on the call January 26, an Object has a
>>>>> 
>>>>> location but it can change over time (in which case it is a NON-FIXED POI),
>>>>> 
>>>>> or not moving (in which case it is a FIXED POI).
>>>>> 
>>>>> 
>>>>> 
>>>>> I welcome feedback.
>>>>> 
>>>>> 
>>>>> 
>>>>> Regards,
>>>>> 
>>>>> 
>>>>> 
>>>>> Christine
>>>>> 
>>>>> 
>>>>> 
>>>>> Object Primitive
>>>>> 
>>>>> 
>>>>> 
>>>>> Goal:
>>>>> 
>>>>> Provide a rich and flexible description of an object (aka a thing)
>>>>> 
>>>>> De-couple or isolate the description of an object and from where it is (a
>>>>> 
>>>>> Place of Interest) and other primitives.
>>>>> 
>>>>> An object has one location at a (temporary-duration undefined) specific
>>>>> 
>>>>> point in TIME but does not have one fixed point over time.
>>>>> 
>>>>> 
>>>>> 
>>>>> An Object of Interest can be a parent to other Objects each with its own
>>>>> 
>>>>> description to allow for the representation of complex objects that are the
>>>>> 
>>>>> aggregate of a collection of Objects (a car, boat, or airplane).
>>>>> 
>>>>> It should not be inferred that each of the elements within the object
>>>>> 
>>>>> primitive are not spatially synonymous, but do refer to the same object.
>>>>> 
>>>>> 
>>>>> 
>>>>> High Level Attribution:
>>>>> 
>>>>> 
>>>>> 
>>>>> Object Name
>>>>> 
>>>>> Object's Absolute Location at last known time
>>>>> 
>>>>> 
>>>>> 
>>>>> Why absolute? Could it not be relative to another location?
>>>>> 
>>>>> If an absolute location is known, it could be added.
>>>>> 
>>>>> If a relative location is known, it could also be added.
>>>>> 
>>>>> I would suggest that both are optional but that either, if in the primitive,
>>>>> must have a time stamp.
>>>>> 
>>>>> 
>>>>> 
>>>>> Identification
>>>>> 
>>>>> 
>>>>> 
>>>>> Object's category [living, non-living]
>>>>> 
>>>>> 
>>>>> 
>>>>> What is the (use) case for distinguishing living and non-living objects?
>>>>> 
>>>>> I think many things can be inferred via the living vs. non-living ID. Having
>>>>> an idea, when available, of the object's living/non-living status may help
>>>>> with recognition and tracking algorithms, for example.  Once you know it is
>>>>> living, and then know if it is an animal, then something may recognize the
>>>>> genus. Then, your algorithms will probably detect deformations more often in
>>>>> a specific zone (the head).
>>>>> 
>>>>> non-living materials/objects can also deform (e.g., water, sand) but using
>>>>> different parameters. The more which can be quickly codified the lower the
>>>>> processing requirement, the faster the response times.
>>>>> 
>>>>> By the way, I'm not attached to this/saying it is required, but I suspect it
>>>>> will have utility.
>>>>> 
>>>>> 
>>>>> 
>>>>> Object's Attribution Details
>>>>> 
>>>>> 
>>>>> 
>>>>> Identification
>>>>> 
>>>>>                ID (optional)
>>>>> 
>>>>> Identification System or Service
>>>>> 
>>>>> ID
>>>>> 
>>>>> Associated IDs
>>>>> 
>>>>> 
>>>>> 
>>>>> What is exactly the meaning of an "ID", a "System or service ID" and
>>>>> 
>>>>> an "Assoicated ID".
>>>>> 
>>>>> 
>>>>> 
>>>>> Don't know, really. Perhaps this is where one stores the living v.s
>>>>> non-living, etc These were part of the Location Primitive so I kept them. I
>>>>> think the Associated IDs would be helpful if an object is part of a larger
>>>>> entity (e.g., the component of a computer, or anything else which is
>>>>> composed of multiple objects)
>>>>> 
>>>>> 
>>>>> 
>>>>> Name
>>>>> 
>>>>> Last Updated On : Date/Time  (optional)
>>>>> 
>>>>> Updated By : owner / author (optional)
>>>>> 
>>>>> Use : public, private, restrictions (optional)
>>>>> 
>>>>> Ownership info : owner of all or part of a POI
>>>>> 
>>>>> Cost - each point can cost more for the people who lease it from the layer
>>>>> 
>>>>> owner.
>>>>> 
>>>>> 
>>>>> 
>>>>> This seems to come directly from look-here.biz. What's the use case for
>>>>> that?
>>>>> 
>>>>> You are correct. This was integrated from Roy Davies' post.
>>>>> 
>>>>> Cost can be unknown, known, it can be flexible (more for me that for a
>>>>> member of a club). I think that having information about the value in
>>>>> monetary terms might be important if you are in a shopping application.
>>>>> 
>>>>> 
>>>>> 
>>>>> Status : Active, blocked, deleted (optional)
>>>>> 
>>>>> 
>>>>> 
>>>>> This is usually (if not always) in relation in a certain context. I do
>>>>> 
>>>>> not see how an object of itself can be "active" or "blocked". If it's
>>>>> 
>>>>> deleted, doesn't it just disappear?
>>>>> 
>>>>> Yes, but once again, these were "inherited" from the location primitive and
>>>>> my goal was to keep the two as similar to one another as possible.
>>>>> 
>>>>> 
>>>>> 
>>>>> Trustworthiness : degree of certainty the author has in the accuracy of the
>>>>> 
>>>>> object
>>>>> 
>>>>> 
>>>>> 
>>>>> What makes an object more or less accurate?
>>>>> 
>>>>> Good catch. It should be a relative assessment of the accuracy of the
>>>>> information, not of the object.
>>>>> 
>>>>> 
>>>>> 
>>>>> Category
>>>>> 
>>>>> 
>>>>> 
>>>>> Object (optional)
>>>>> 
>>>>>                                Type
>>>>> 
>>>>> Living - composed of one or more cells
>>>>> 
>>>>> Non-living - inanimate, not composed of cells
>>>>> 
>>>>> 
>>>>> 
>>>>> [Then we can classify according to animal or plants, mammal, etc!]
>>>>> 
>>>>> 
>>>>> 
>>>>> Again, what is the use case for this classification system?
>>>>> 
>>>>> 
>>>>> 
>>>>> Addressed (to the extent I can) above. If you are going to assign a business
>>>>> code to places (e.g., the place is commercial, this kind of a commerce, etc)
>>>>> then, why not have taxonomy in the sense of "nature"? Just thinking outside
>>>>> the box!!
>>>>> 
>>>>> MORE OPTIONAL INFO ABOUT OBJECTS:
>>>>> 
>>>>> Circumference/radius (the description applies to this Object plus the space
>>>>> 
>>>>> around it?)
>>>>> 
>>>>> 
>>>>> 
>>>>> Isn't this covered in Location? An object is linked to a Location,
>>>>> 
>>>>> which defines its spatial boundaries...
>>>>> 
>>>>> Not necessarily. If the object is not linked to a location, then this rule
>>>>> does not hold. And the POI might refer to simply the size, if it is known,
>>>>> in order to aid with the recognition of how far away it is from the user,
>>>>> how to display the digital data in relation to it, etc.
>>>>> 
>>>>> 
>>>>> 
>>>>> Is it planar or 3D?
>>>>> 
>>>>> If it is planar, what are its X&Y dimensions? This is the equivalent of area
>>>>> 
>>>>> for the location primitive.
>>>>> 
>>>>> 
>>>>> 
>>>>> If it is 3D, what is known about it?
>>>>> 
>>>>> Volume
>>>>> 
>>>>> Density
>>>>> 
>>>>> 
>>>>> 
>>>>> Identification (optional)
>>>>> 
>>>>> Supplier: who made this object?
>>>>> 
>>>>> Version: what is its version?
>>>>> 
>>>>> Associated Object ID
>>>>> 
>>>>> Trustworthiness of this information?
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> What are its relationships to other objects?
>>>>> 
>>>>> Is it near a fixed object? How near?
>>>>> 
>>>>> Does it belong to someone? A company or an individual?
>>>>> 
>>>>> Connected- Part of a larger entity (a motor?)
>>>>> 
>>>>> Independent- not part of a larger entity
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> ________________________________
>>>>> 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.
>>>> 
>>>> 
>>>> 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.
>>>> 
>>> 
>>> 
>>> 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.
>>> 
>> 
>

Received on Tuesday, 15 February 2011 10:24:31 UTC