SIGIA-L Mail Archives: SIGIA-L: UML for Information Architectur
SIGIA-L: UML for Information Architecture: an interesting idea!
From: chuck.lutz_at_telelogic.com
Date: Wed Jun 28 2000 - 11:15:52 EDT
Hi folks,
This is my first post to SIGIA. I've sometimes been found opining
over on InfoDesign-Cafe.... Anyhow, this topic piqued my interest.
The evolution of "the web" from serving static HTML documents
to delivering today's dynamic, full-scale applications clearly
brings the task of "creating the web" much closer to software
engineering. In its current state, web creation represents
perhaps the tightest coupling of right-brain/left-brain stuff
we've ever seen:
It used to be that a graphic designer could pass him/herself off
as a "new media guru" just by learning a little HTML, but as more
functionality and dynamism has crept in I think these folks have
had a harder time. I think many use borrowed libraries of e.g.
Javascript but I strongly doubt many understand it (I know of many
instances of this practice).
On the other hand, the world seems to be full of design-starved
web sites hacked together by propellerheads with no interest in
usability, good visual design, IA, you name it. They seem not to
understand that "good" code has no impact on the user/customer
experience.
I'm interested in investigating and _furthering_ the merge between
information design and software engineering. Toward that end, I
think about how what is known in SE can be applied to ID and vice
versa.
Regarding the choice of a notation for IA use, someone suggested
UML as a possibility. I found this interesting because we're start-
ing to see instances of UML use in web app design in the field. How-
ever, I wonder how much IA stuff has been considered thus far. I bet
it could be used rather well.
--I declare the fact that I work for a Rational-alternative UML
tools vendor now, in the interest of full disclosure! :-) --
Lawrence Lipkin wrote:
>The sites that we make are always dynamically served. We're starting to see
>lots of variables from business logic, personalization, etc. determining
>what is served within our page templates.
I think the thing to borrow here from the world of object-oriented s/w
engineering is "classes" vs. "objects"/"instances". The idea of dynamically
synthesized web pages fits in well here, especially in the increasingly
popular model of a "framework" of web page templates fleshed out on-the-fly
with content from a "content repository" of some sort. For example, you may
have 56 "product" pages in a subweb of "our products". Each individual prod-
cut is a member of the class Product. Also, you can sort-of think of the
relationship between the "Our Products" page and the Class of "products"
as being a Generalization. The details of the "content" bits that fill in
specific areas of the templates can be thought of as specifying the "attrib-
utes" of each Object in the Class "Products". This will only be understood
by OO-familiar folks but I don't have room to explain it here; besides I
ramble on for too long as it is.
Now that XML is here, I wonder if the mapping of "content bits" to "Class
attributes" is made more conceptually straightforward.
>At some point a "page" is really just a contextual instance of behavior
>determined by a series of components on a server.
I tried to say this another way above.
The benefit is, I think, from an Information _Architecture_ POV: You don't
have to worry about all the possible instances of every combination of
content bits. Besides, that is not the point of architectural descriptions.
You want a "high level" view of the site structure. Such a view can be
refined to give more detail as needed (part of the beauty of the object-
oriented approach and UML). Typically in s/w engineering, you want a static
"architectural" view and a dynamic "behavioral" view. I don't know much
about the details of web site dynamics (don't know much about JavaScript!)
but I would venture to guess that they could be modeled well with UML's
Statechart notation.
>The high level sitemaps that we make have become more like filmic
>storyboards for describing a user experience to our clients. They are an
>extremely necessary step in the design process -- but always trending
>towards abstraction.
...that's exactly the point. Yes; extremely necessary in order to start
defining the architecture that won't change. However, we abstract so as
not to get bogged down in as-of-yet unneccesary (or more likely, UNKNOWN!)
details.
Sometimes the architecture changes dynamically too. There are ways to
model that.
>The abstraction is inadequate to describe all of the possible instances of
>user experience -- but very good for talking about what is most important.
The problem I find in trying to discuss this with people is that I either
find propellerheads or artists, but not so many good combinations of both.
Admittedly, I'm more of the former. I live vicariously through my wife,
who is trained in art and architecture but has programmed for fun since she
was 10. Although my formal training is in electrical engineering and s/w,
I'm quite interested in IA, usability, graphic design, document design, the
works. It is all vitally important for successful software experiences.
The sooner the software industry learns this, the better we all will be.
Here's an instance of UML use in web design:
http://cseng.aw.com/bookdetail.qry?ISBN=0-201-61577-0&ptype=0
I haven't checked it out yet but I doubt I'd find IA modelling. This web
page gives me the impression that the book is written from a pure s/w POV.
Regards,
Chuck Lutz
This archive was generated by hypermail 2.1.2
: Sun Nov 23 2003 - 22:54:20 EST
|
