ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Watchout Watson: Here comes Amazon Machine Learning

To: "'[ontolog-forum] '" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "Matthew West" <dr.matthew.west@xxxxxxxxx>
Date: Sun, 3 May 2015 08:19:53 +0100
Message-id: <01ac01d08571$8cfb9140$a6f2b3c0$@gmail.com>

Dear Thomas,

This clarification is helpful.

 

Rows in most implicitly temporal (i.e. apparently non-temporal) relational tables represent contingent states of affairs, e.g. that Customer 123 currently has name = "Smith" and status = "Platinum". 

[MW>] The current state of affairs.

 

Rows in unitemporal tables include the open or closed time period during which C123 had that name and that status. 

[MW>] This is the full proposition under discussion then.

 

Rows in bitemporal tables include that first time period (valid time, called "state time" in my recent book). But they also include a second time period, which represents the time from when the row was physically created to when it was logically deleted (i.e. marked deleted, but not physically deleted).

[MW>] OK. This is data about the record itself and its status.

 

Rows in tritemporal tables -- discussed in my book but not in the current literature -- include a third time period which represents when a person or group of persons (such as the enterprise that owns the database in question) was, is or will be willing to assert that the statement made by a row correctly describes a state of affairs, i.e. is a true statement. 

[MW>] OK. So when a record representing me as an employee is included in a company database, it includes my date of birth as the start date of when I existed, the record creation date as the date the record about me was created, and the date my application form was received by the company (or perhaps checked by them) as the date they accepted it as true. When they later find I had lied about my date of birth (perhaps because they have an ageist employment policy) they add the date they discovered this and logically delete the record when the admin get round to it, and perhaps create a new record with correct data and fire me.

Sounds good to me.

 

Regards

 

Matthew West                           

Information  Junction

Mobile: +44 750 3385279

Skype: dr.matthew.west

matthew.west@xxxxxxxxxxxxxxxxxxxxxxxxx

http://www.informationjunction.co.uk/

https://www.matthew-west.org.uk/

This email originates from Information Junction Ltd. Registered in England and Wales No. 6632177.

Registered office: 8 Ennismore Close, Letchworth Garden City, Hertfordshire, SG6 2SU.

 

 

 

 

Updates to these tables withdraw currently asserted rows with open time periods (most of them) because an update makes such rows false. A current open time period row might say, for example, that C123's status is Platinum from some past point in time into the indefinite future. But with the update, we know that statement is false, and that its open time period must be closed as of the time of the update.

 

Your description of fluents, at first glance at least, seems to describes rows in tables which contain both the data (C123 has name="Smith" and status="Platinum"), and also a description of what the world must be like in order to make that statement true. But since what the world must be like in order to make that statement true is precisely that the designated customer has name="Smith" and status="Platinum",. it doesn't seem that a description of "what the world must be like in order to make (a) statement true" adds any information to the original row.

 

Which suggests that I must currently fail to understand what you mean by "fluents". Can you enlighten me?

 

Tom

 

 

 

 

On Friday, May 1, 2015 5:09 PM, Rich Cooper <metasemantics@xxxxxxxxxxxxxxxxxxxxxx> wrote:

 

Dear Matthew, Thomas, John, ...

 

TJ> I have proposed a third temporal dimension -- that in which the statements made by rows in tables are asserted to be true. For example, a logically deleted row remains in a database as one which no longer represents the assertion of the statement it makes. And just as we can utter/write statements (de-indexicalized declarative sentences) that, starting at some point in time, we are no longer willing to assert represent statements we believe are true, so too we can enter rows into database tables that represent statements that we are not yet, at that time, willing to assert represent true statements. The former rows are tagged with "speech act" time periods (see below) that are entirely in the past, and the latter with time periods that are entirely in the future. Again, this is not in the current SQL standard, nor in the current comp sci literature.

[MW>] I’m still not clear (but interested).

 

RC> I think you are describing what is best represented as fluents in the table.  A "fluent" row comprises the predicate's specified value of true or false, and its parametric bindings to objects and properties.  But the fluent row also stores the truth value that this row currently represents - i.e., true or false.  For example:

 

_a^2 + _b^2 = _c^2

 

is the 90 degree triangle Pythagorean constraint on the length of the three sides.  To determine if the Pythagorean constraint is true, we need values for _a, _b and _c to check the equation.  But we can also store the truth value _tf with the value of a given triangle we wish to store in the knowledge base.  So the database record would be:

 

   < _tf : Boolean;  _a : number;  b: number;  c : number >

 

Then, given instance values for a, b and c to ground _a, _b and _c, the Pythagorean constraint on that instance would be evaluated true or false.  That evaluation, compared with the stored value _tf, would drive an application's interpreted decision to assert that the instance is Pythagorean if and only if the evaluation is equal to _tf, which in this example must be stored as true in the row's value for _tf. 

 

Sincerely,

Rich Cooper,

Rich Cooper,

 

Chief Technology Officer,

MetaSemantics Corporation

MetaSemantics AT EnglishLogicKernel DOT com

( 9 4 9 ) 5 2 5-5 7 1 2

 

From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Matthew West
Sent: Friday, May 01, 2015 1:37 PM
To: 'Thomas Johnston'; '[ontolog-forum] '
Subject: Re: [ontolog-forum] Watchout Watson: Here comes Amazon Machine Learning - ZDNet- 2015.04.10

 

Dear Thomas,

 

John, Matthew:

 

Commercial databases have no explicit ontology beyond the ontological commitments made in their database catalogs. I try to articulate an implicit ontology which I believe they all share, in my book that John alluded to.

 

The current SQL standard recognizes valid time and transaction time, the former being the time in which the things and objects the database is about exist, and the latter being the time in which rows reflecting these things and changes are entered into the database and/or marked as logically deleted.

[MW>] Do you mean by “logically deleted” that the statement is no longer true but once was, or that they were never true (say the record was entered in error or a change of understanding)?

 

I have proposed a third temporal dimension -- that in which the statements made by rows in tables are asserted to be true. For example, a logically deleted row remains in a database as one which no longer represents the assertion of the statement it makes. And just as we can utter/write statements (de-indexicalized declarative sentences) that, starting at some point in time, we are no longer willing to assert represent statements we believe are true, so too we can enter rows into database tables that represent statements that we are not yet, at that time, willing to assert represent true statements. The former rows are tagged with "speech act" time periods (see below) that are entirely in the past, and the latter with time periods that are entirely in the future. Again, this is not in the current SQL standard, nor in the current comp sci literature.

[MW>] I’m still not clear (but interested).

 

Regards

 

Matthew West                           

Information  Junction

Mobile: +44 750 3385279

Skype: dr.matthew.west

This email originates from Information Junction Ltd. Registered in England and Wales No. 6632177.

Registered office: 8 Ennismore Close, Letchworth Garden City, Hertfordshire, SG6 2SU.

 

 

Assertions (and their withdrawals) are speech acts, and so in my book, I attempt to introduce this third temporal dimension as that. So I distinguish (i) rows in tables as physical inscriptions; (ii) statements made by those rows (semantics); (iii) speech acts associated with those semantics such as assertions and withdrawals, also assents, dissents, and the whole panoply of propositional attitudes (pragmatics); and (iv) the propositions expressed by those statements (the Holy Grail of why we manage data in the first place).

 

I think that all of these are managed objects which can be represented as managed objects in relational databases. The only managed objects in today's databases, on the other hand, are the physical tables, rows and columns themselves. Constraints preserve an imperfect mapping to an interpretation, and the understanding of that interpretation is a matter that falls outside the scope of the management of data itself. Or, at least, so I understand a basic assumption underlying the current state of affairs -- an assumption I do not make myself.

 

A bit of a soapbox email here. Apologies, if needed.

 

Tom

 

 

On Wednesday, April 29, 2015 8:13 AM, John F Sowa <sowa@xxxxxxxxxxx> wrote:

 

Dear Matthew,

Possibility and necessity affect the logic, not the ontology:

> Another problematic category is possibilia (things that might be,
> or possibly are in some parallel universe).

They can be treated in the same way as plans for the future.
For example, if you're designing an airplane or a bridge, it's
a possibility until it's actually built.

> The criteria for including possibilia (or not) is utility vs
> the baggage that comes with the extra commitment.

The categories of parts, part numbers, etc., might be empty
in actuality, but they are specified in the ontology by the
same methods before and after the things are built.

There are, of course, issues about storing information about the
future in the database -- orders for future delivery of things
that don't yet exist, reservations for hotels, travel, etc.
The orders and reservations exist in the present (or past),
but they refer to things and events in the future.

Tom Johnston wrote a book about time and temporal issues in
databases.  Perhaps he might care to comment on this point.

Following is an article in which I discuss issues about
modality, possible worlds, and the laws that govern them:

    http://www.jfsowa.com/pubs/worlds.pdf
    Worlds, models, and descriptions

And by the way, possibilities are another area where a strictly
nominalist position (e.g., Quine's or Goodman's) gets into trouble.

Clarence Irving Lewis, who defined the first modern versions
of modal logic, had been the chair of the philosophy department
at Harvard while Quine was a student and later a professor.

But Quine was very strongly opposed to any version of modal
logic and any talk of possibilia.  Hao Wang, who had earned
a PhD under Quine's supervision, was very critical of Quine's
attitude.  He called it *logical negativism*.  See

    Wang, Hao (1986) Beyond Analytic Philosophy: Doing Justice
    to What We Know, MIT Press, Cambridge, MA.

 

 


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)

<Prev in Thread] Current Thread [Next in Thread>