ftp.cs.arizona.edu/ftp.cs.arizona.edu.tar/ftp.cs.arizona.edu/tsql/doc/tdbglossary.mail/[e-mail] _Fri Aug 13 22:40:44 1993.msg

Internet Message Format | 1996-01-31 | 13KB

Received: from iesd.auc.dk by optima.CS.Arizona.EDU (5.65c/15) via SMTP
    id AA00396; Fri, 13 Aug 1993 13:42:06 MST
Received: from yellow.iesd.auc.dk by iesd.auc.dk with SMTP id AA03916
  (5.65c8/IDA-1.5/MD for <[e-mail]>); Fri, 13 Aug 1993 22:39:57 +0200
Received: by yellow.iesd.auc.dk id AA03614
  (5.65c8/IDA-CLIENT/MD for [e-mail]); Fri, 13 Aug 1993 22:40:44 +0200
Date: Fri, 13 Aug 1993 22:40:44 +0200
From: "Christian S. Jensen" <[e-mail]>
Message-Id: <[e-mail]>
To: [e-mail]
Subject: Status and Proposals

Subject: Status and Proposals
To: [e-mail]
From: [e-mail], on behalf of the glossary editorial board

In our efforts to consolidate the glossary on temporal database
concepts, we have so far established an overall structure for the
glossary. Nine days ago, the editorial board proposed a categorization
of glossary entries and requested comments from the community. A final
categorization will be defined soon, based on the resulting feedback.
That is the final major step in consolidating the structure of the
glossary.

Also nine days ago, the community was asked to contribute to the task
of consolidating the individual glossary entries. The deadline for
proposals for improvements is today (Aug 13). Below are twenty
proposals from the editorial board. These proposals are based on
feedback from the community, at the workshop and after. Each proposal
is concrete and concerns the modification or omission of individual,
identified glossary entries.

Any comments on the proposals are due on Tuesday, August 17. Recall
that the deadline for completing the entire glossary is Monday, August
23, 1993.

***Proposal 1:
Entry: valid time
New entry: Append the following to the current definition.

"Valid times are usually supplied manually by the user."

Comment: This is a simple addition, prompted by comments by Nikos
Lorentzos and Yannis Mitsopoulos.

***Proposal 2:
Entry: transaction time
New entry: Replace the sentence "Transaction times may be
implemented using transaction commit times." by

"Transaction times may be implemented using transaction commit times,
and are system-generated and -supplied."

Append "transaction commit time" to the list of previously used names.

Append the following to the discussion of naming section.

"Transaction commit time is lengthy ($-$E2), but more importantly, the
name appears to indicate that the transaction time associated with a
fact must be identical to the time when that fact is committed to the
database, which is an unnecessary restriction and is imprecise ($-$E8,
$-$E9). It is imprecise because the transaction time of a fact in
general is a transaction-time element, not a single time instant as
implied."

Comment: Part of this proposal was prompted by comments by Nikos
Lorentzos and Yannis Mitsopoulos. It is a relatively uncomplicated
addition of material.

Proposal 3:
Entry: transaction-timeslice operator
New entry: Replace the sentence (in the definition) "It also takes as
argument a time value not exceeding the current time, {\em NOW\/}." by

"It also takes as an argument a time value---a chronon in discrete
models and a time instant in continuous models---not exceeding the
current time, {\em NOW\/}."

Comment: This proposal is intended to make what is meant by "time
value" precise. I.e., here a time value is *not* a an interval or an
element.

Proposal 4:
Entry: valid-timeslice operator
New entry: Replace the sentence (in the definition) "It takes as
argument a time value." by

"It takes as argument a chronon (discrete models) or a time instant
(continuous models)."

Comment: This proposal is intended to make what is meant by "time
value" precise. I.e., here a time value is *not* a an interval or an
element.

Proposal 5:
Entry: chronon
New entry: Replace the current definition by the following:

"In a data model, a one-dimensional {\em chronon} is a
non-decomposable time interval of some fixed, minimal duration.
An $n$-dimensional chronon is a non-decomposable region in
$n$-dimensional time. Important special types of chronons include
valid-time, transaction-time, and bitemporal chronons."

Add an explanation part with the following prose.

"Data models that support a discrete notion of time may represent a
time line by a sequence of non-decomposable, consecutive intervals of
identical duration. These intervals are termed chronons. A data model
will typically leave the particular chronon duration unspecified, to
be fixed by a particular implementation of the data model."

Comment: The current definition defines a chronon as a duration, which
is a mistake.

Proposal 6:
Entry: interval
New entry: Change the name to "Time Interval." Change the definition
to the following.

"A {\em time interval} is the time between two instants. It may be
represented by a set of contiguous chronons."

Change the discussion part to this:

"The name ``time interval'' (``interval'' when ``time'' is clear from
the context) is widely accepted (+E3). The name ``period'' often
implies a cyclic or recurrent phenomenon ($-$E8, $-$E9)."

Comment: This proposal integrates a proposal by Gio Wiederhold and
Pernici et al. with comments by Nikos and Yannis.

Proposal 7:
Entry: Time-invariant Attribute

Do not make this entry part of the current glossary. The definition is
very imprecise and should be improved significantly before the entry
is adopted. If the entry is included as it is, it is very likely to
change in future versions of the glossary.

Comment: We claim that only well-defined concepts are included in the
glossary. This entry seems to be too imprecise.

Proposal 8:
Entry: Time-varying Attribute

Do not make this entry part of the current glossary. The definition is
very imprecise and should be improved significantly before the entry
is adopted. If the entry is included as it is, it is very likely to
change in future versions of the glossary.

Comment: We claim that only well-defined concepts are included in the
glossary. This entry seems to be too imprecise.

Proposal 9:
Entry: fixed span
New entry: Replace the existing definition by this:

"A span is {\em fixed\/} if it possesses the special property that
it's duration is independent of the context."

Add an explanation section with these contents:

"As an example of a fixed span, ``one hour'' always, independently of
the context, has a duration of 60 minutes. To see that not all spans
are fixed, consider ``one month,'' an example of a variable span. The
duration of this span may be any of 28, 29, 30, and 31 days, depending
on the context."

Comment: The current definition defines the notion of context-dependent
and independent durations in the first sentence. The new definition
for fixed span defines a fixed span right away, and the optional
explanation part is utilized.

Proposal 10:
Entry: variable span
New entry: Replace the existing definition by this:

"A span is {\em variable\/} if it's duration is dependent on the
context."

Add an explanation section with these contents:

"Any span is either a fixed span or a variable span. The obvious
example of a variable span is ``one month,'' the duration of which may
be any of 28, 29, 30, and 31 days, depending on the context. The span
``one hour'' is fixed because it always, independently of the context,
has a duration of 60 minutes.

The presence of variable spans represents a (necessary) complication
of temporal support that has no obvious parallel in, e.g., spatial
support."

Comment: This change is proposed to obtain consistency between the
entries for fixed and variable spans.

Proposal 11:
Entry: initiation
New entry: Use this definition:

"The distinguished value {\em initiation,\/} associated with a
relation, denotes the time instant when a relation was created.
``Initiation'' is a value in the domain of transaction times and has
no valid-time semantics.

Comment: The current definition seems to indicate that the time when a
database is created and the time when the first update takes place are
identical. I do not think this is always true. Also, different
relations may have different initiation values.

Proposal 12:
Entry: calendar
New entry: In the discussion part, replace the third word "generally"
by "most often."

Comment: The proposal is prompted by a comment by Nikos and Yannis.
"Generally" could be confused with "always," "most often" cannot.

Proposal 13:
Entry: temporal value integrity
Entry: time sequence
Entry: history-oriented

These three proposals exhibit a significant overlap and should not be
part of the present glossary. A single, integrated proposal that
captures the underlying concept may be included instead.

Comment: The feeling at the workshop was that integration of these
entries into the glossary was premature. It was felt that the
underlying concept was perhaps not sufficiently well-defined,
well-understood, and widely used (three of the four criteria for
inclusion of concepts).

Proposal 14:
Entry: temporal rule

This entry must be revised before it can be included. Part of the
contents of the discussion section belongs in the new, optional
explanation section, and the precision of the definition should be
improved (e.g., (1) "time point" is not defined, (2) why are only
points and temporal elements and not intervals mentioned, (3) is the
notion of time restricted to valid or transaction time (as in
"temporal database") or does also user-defined time count)

Comment: The proposal is motivated by some imprecisions that were
identified at the workshop.

Proposal 15:
Entry: temporal interpolation

This entry depends on the notion of time sequence, so if "time
sequence" is not part of the current glossary, this entry must be
rewritten before it can be part of the glossary. "Time point" is also
used. Should there be an entry for "time point"?

Comment: Motivated by necessary consolidation requirements.

Proposal 16:
Entry: period of indeterminacy
Entry: admissibility interval

These two entries have conflicting names. The conflict must be
resolved before the concept can be added to the glossary.

Comments: Two conflicting entries should be in the glossary.

Proposal 17:
Entry: interval relation

This entry is incomplete, and it names a concept which has little to
do with what is usually thought of as a relation in temporal
databases. Inclusion is premature until the entry is renamed and
expanded.

Comment: Renaming was recommended at the workshop; Making the entry
complete is dictated by the need for consistency.

Proposal 18:
Entry: event occurrence time
Entry: interval relation holding time

The former concept denotes the time when an event is true; the latter
denotes the time when an interval relation is true. Why not
"occurrence time" or "holding time" in both cases. Also, "interval
relation holding time" sounds as if we are dealing with some interval
relation that holds time, which is a bit confusing! If "interval
relation" is an unfortunate name, so is "interval relation holding
time." Thus, these proposals should not be considered until the next
glossary version.

Comment: The proposal is motivated by a need for consistency.

Proposal 19:
Entry: temporal qualification
New entry: Use this definition:

"The {\em temporal qualification\/} of a statement is the component of
the statement that specifies the temporal localization of the denoted
fact. A temporal qualification may be expressed in terms of instants,
chronons, intervals, elements, and spans."

Comment: Before, elements were (unintentionally!?) excluded from
consideration.

Proposal 20:
Entry: instant/event
New entry:

\entry{Instant}{instant}

\epart{Definition}

An {\em instant\/} is a time point of infinitesimal duration.

\epart{Explanation}

The concept of an instant derives from a perception of time as being
continuous. The time domain is perceived to be isomorphic to the real
numbers. Thus, an instant is the temporal counterpart of a real
number.

In a data model that supports a time line using chronons (isomorphic
to the natural numbers or a subset thereof), an instant is represented
by a chronon. A single chronon represents multiple instants.

\epart{Previously Used Names}

Event, moment.

\epart{Discussion of Naming}

``Event'' is already used widely within temporal databases, but is
often given a different meaning (+E3, $-$E5), while ``moment'' may be
confused with the distinct term ``chronon'' ($-$E7).

Comment: This proposal is based on the recommendation by the workshop
and on proposals by Wiederhold and Pernici et al.