At 12:30 13.07.98 GMT0BST, you wrote:
>> From:          Thomas Muhr - Scientific Software Development
<[log in to unmask]>
>>... stuff deleted
>> 2. You can use hypertext networks of linked data segments including text,
>> graphic segments and audio data (video in the future) to not only navigate
>> between the bits and pieces, but also to represent macro structures (story
>> lines etc).
>> 3. Source and target objects are not "points"  but segments, embedded in
>> their context, so you have a lower chance to "get lost in hyperspace". In
>> addition, you can traverse to the more abstract level of codes and back to
>> the data.
>I'm not sure I understand the distinction that you're making 
>here.    As I understand it, in a hypertext you have the ability to 
>create the relevant structures from the ground up as it were.  Frex <--??
>ATLAS/ti has several object types (PDs, quotes, memos and others).  
>In a hypertext then one could presumaly emulate this structure by 
>conceptualizing every node to be one of the ATLAS/ti objects.  I 
>suppose the question would then be: why?
Such conceptualization (not many HTS which allow this) would only make sense
if different "procedural" semantics could be attached to such categories, as
is the case in ATLAS.ti. So, even with data segments, codes, memos, etc.
displayed in a general network, they still have different semantics. First
of all, primary documents and data segments are here the "interpreted" while
codes and memos are outcomes of the interpretative process. The latter
"interpretations" of course can be multi level in that some codes become
themselves the focus of the interpretative process. This is the essence of
the "conceptual work phase" supported by ATLAS.ti.
Different procedures associated with the object types include different
click-actions, specialized context-menus, and different sets of relations
offered for linking objects. Quotation-quotation (hypertext) links and
code-code (semantic) links use different sets.
I think it is important that systems are able to keep these entities in
different "leagues", as they support different epistemological aims in the
process.
Have I answered you remark correctly or were you having the distinction
between representation and navigation in mind?
>
>One of the purported advantages of hypertext is that allows 
>associative exploration.  Coding structures appear to act quite 
>differently in that they bounce you from data through code to a 
>selection list of other bits of data grouped around that code.  It 
>From a navigational standpoint, exactly. Formally, codes (precisely: first
order codes, which are directly associated with the data) are nothing else
but initially unordered sets of data bits with retrieval operations
constrained to set or Boolean operations based on "equality". This is where
I think hypertextual structure preserves more of the "story line", being
able to represent something like "Segment A contradicts segment B", which is
at least a bit cumbersome with codes alone. 
>appears to be more like using a library catalogue.  Although it can 
>be emulated in a hypertext it would seem to make more sense to use 
>hypertext for what its best at and if I want to set-up a coding 
>structure then use a program oriented around it.
>
>> 4. Semantic and hypertext networks can be mixed showing the interrelation
>> between code and data level quite densely.
>> 5. Traversal between the data segments is supported by displaying "hot"
>> links in the margin area alongside the segmented text.
>> 
>> One of the reasons to include hypertext functionality was the inability of
>> purely code-based systems to express statements like "argument x is in
>> contradiction with argument y".
>Yup.
> 
>> I would be interested in your ideas of what an architecture of a "full
>> fledged hypermedia systems" should include (besides nicer buttons...).
>Part of my problem is that this is part of the problem :-)  
Sic!
>Seriously, there are at least two facets of the problem important to 
>me: the use of hypertext as a data-explorer and the use of hypertext 
>as a presentational tool.  In an ideal world what I would like is to 
>be able to use a hypertext system as an aid to exploring and 
>analyzing the data and then, in that same system, to take that work 
>and present it in a manner understandable to another person.  In a 
>sense there would be a seamless progression from analysis to 
>presentation (in a sense one could argue that part of analysis 
>consists of repeated attempts to present your data to yourself in an 
>attempt to make sense of it thus it is only a small step, maybe..., 
>from making sense of the data to yourself and then making sense of 
>the data for others.)
The qualitative researcher as an "information architect building an
n-dimensional informational space" ? The idea of offering HTML generation in
ATLAS.ti is an attempt to provide a structured representation (of course not
as accessible as ATLAS.ti itself) to collaborators and even laymen
(concerning ATLAS.ti). There is a lot to be done there, of course. Same goes
for the input end to process "rich" qualitative data and models from other
projects and tools via XML or other tagged formats.

>
>In terms of qualitative research then I think I'm looking for a 
>system that:
>1) handles a range of media (text, graphics, audio, video)
>2) supports sophisticated hyperlinking between these media.  In 
>essence this requires at least the following:
>	i) link categorization, e.g. the ability to name links
Done.
>	ii) bi-directional linking, e.g. a "return" button
Already there.
>	iii) many-to-one and one-to-many linking.
Done.
>	iv) "text" to node linking and vice-versa.  E.g. the ability in a 
Done.
>video file to link to a set of frames within that video and, also, to 
>set up buttons that are clickable based on the currently playing 
>video frames.
>3) Ability to have multiple windows open.
Done.
>4) Ability to provide graphical overviews of the complete hypertext 
>or subparts of it.
Done
>5) Sophisticated search/query engines.
Well, improvable.....
>
>There are then the "dream" bits such as ability to launch other 
>programs (e.g. ability to intergrate with the internet or an 
>intranet), collaborative work facilities and so on.
Partly.

I found a cute program that does the latter quite nicely, called The Brain
(don't have the URL around).

>
>Of course no such program exists.  Of the CAQDAS packages ATLAS/ti 
>seems closest and of the extant hypertext packages StorySpace seems 
>the most useful.  However nether supports video and given that the 

Have you considered knowledge acquisition systems as well? I am not informed
about the current state-of-art. By the way, a KA system influenced the
development of ATLAS.ti (KADS).

>majority of our data is likely to be video this is somewhat of a 
>major problem.  There are sophisticated hypermedia packages out there 
>(e..g. Director and AuthorWare) but they're based around slick 
>presentation and seem useless for theory building.  An initial 
>decision to try to use AuthorWare and hack it into something useful 
>for data exploring is failing horribly because of its relative 
>poverty of hypertext functionality so we are at a bit of a quandry 
>here.  Hence my email.

There are some hairy topics with clean semantics concerning time based data
in general and I am not really happy yet with the basics. However, we will
be integrating two level segmented video data (frame sequence plus frame
region).

So much for a new edition on "science and commerce"!-)

Regards
- Thomas

>
>---Bruce
___________________________________________________________________
                             Dipl.-Psych. Dipl.-Inform. Thomas Muhr
Scientific Software Development  -  Internet: http://www.atlasti.de    
e-mail: [log in to unmask]  TEL [+49 30] 861 14 15  FAX 864 20 380
ATLAS-TI mailing list (join, leave):   www.atlasti.de/joinlist.html



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%