RE: [unrev-II] Jack's Use Case: Context-sensitive representation

From: Regev, Gil (gil.regev@epfl.ch)
Date: Thu Nov 02 2000 - 08:57:59 PST

  • Next message: Jack Park: "[unrev-II] Towards a summary of documents"

    Eric,
     
    check out the Topic Maps trail at http://www.topicmaps.com/
    <http://www.topicmaps.com/> I believe they have defined roles of
    associations the same way as you do in this message. The trail starts at
    http://www.topicmaps.com/machinery/intermediary.asp?file=../content/trai
    ls/resolved\step1.xml
    <http://www.topicmaps.com/machinery/intermediary.asp?file=../content/tra
    ils/resolved\step1.xml> they start discussing roles of associations at
    step 7. You can find the ISO standard at
    http://www.y12.doe.gov/sgml/sc34/document/0129.pdf
    <http://www.y12.doe.gov/sgml/sc34/document/0129.pdf>
     
    Gil
     

    -----Original Message-----
    From: Eric Armstrong [mailto:eric.armstrong@eng.sun.com]
    Sent: jeudi, 2. novembre 2000 00:54
    To: unrev-II@egroups.com
    Subject: Re: [unrev-II] Jack's Use Case: Context-sensitive
    representation

    Thanks again for raising this critical question,
    Jack. It got the juices flowing, and this morning
    the answer arrived, as it always does, like a
    tiny bolt of lightning in the quiet of morning.

    There is a analagous concept in O-O design: the
    concept of a *role*. In that context, the idea
    of a role emerges when one realizes that a particular
    noun in a system description does not represent a
    new object, but rather the name of a variable -- a
    particular way of characterizing an existing object.

    Thinking in terms of "roles" led me to put your
    question together with Doug's magnificent
    tour de force: the concept of a adding
    *view control* to a link.

    In essence, the situation with Sally is that there
    are multiple views. "Sally" can be representated as
    an object with many views (if she is thought of as
    a hierarchy or graph of characteristics) or as an
    object with many interfaces (if she is thought of
    as a classic O-O object).

    The important point is that "referencing" the
    object known as "Sally" always takes place in
    some *context*, and in that context, one view
    or another is appropriate. That "view" of her
    characterisitcs constitutes the "role she plays".

    [Intentional ambiguity alert: You "play a role"
    when you are on stage, in the sense that it is one
    view of who you are. You also "play a role" in
    any given context, in the sense that you fulfill
    some specific set of functions.]

    The moral, then, is that any link to any object
    in the system must be qualifiable with respect
    to it's "role", in the same way that any link
    to any part of hierarchy must be qualifiable
    with respect to view.

    For me, the moral is that any reference to a
    globally unique user ID must also be qualifiable
    with respect to a "role" or "view" of that
    user.

    The reason: User IDs are used for attribution.
    Every attribution should become a link to
    information about that individual.

    But the information displayed for that individual
    should depend on the context. The "bio" paragraph
    I use for technical articles is significantly
    different from the one I use for exercise/nutrition
    articles. In the same way, the view of "me"
    presented by traversing a user ID link should be
    different, depending on the context from which it
    came.

    eGroups Sponsor
     <http://click.egroups.com/1/10062/5/_/444287/_/973122872/>

     
    <http://adimg.egroups.com/img/10062/5/_/444287/_/973122872/_Y!-photos-2.
    gif>

    Community email addresses:
      Post message: unrev-II@onelist.com
      Subscribe: unrev-II-subscribe@onelist.com
      Unsubscribe: unrev-II-unsubscribe@onelist.com
      List owner: unrev-II-owner@onelist.com

    Shortcut URL to this page:
      http://www.onelist.com/community/unrev-II
    <http://www.onelist.com/community/unrev-II>



    This archive was generated by hypermail 2b29 : Thu Nov 02 2000 - 09:35:32 PST