Lynda Hardman accepts the blame. Intellectual contributions by Jacco van Ossenbruggen and Oscar Rosell.
This bluenote is trying to sort out our requirements for a hypermedia communicative device model. Basically we don't really know what a communicative device is, but hopefully by the end of writing this things will be clearer. In the context of the layers in the Cuypers system, this model should describe the abstraction level between the the semantic/rhetorical level and the hypermedia formatting object tree level.
Please note that my use of communicative device should not be confused with any previous uses of the term within the group. Perhaps the term should be changed, but for now I am aiming at a shared understanding of what we are talking about, and then we can reconsider the name.
A communicative device is a level of abstraction above that of the HFO and below that of a discourse representation. I don't want to get into what a discourse representation may or should be, but assume that it exists and contains sufficient information for transforming it into one or more communicative devices which, when presented to the viewer, will successfully communicate the message encoded in the discourse structure. The communicative device is above the level of the HFO. The HFO talks about things in the "output" world (pages in books, screens on computers) and communicative devices (CDs from now on) are at the "document level" and are similar to scenes in films and chapters in books. They are neither output oriented nor semantic oriented.
To start the ball rolling, here are some communicative device names, and a description. (Maybe I can improve these when I have been through the rest of the document but, for now, bear with me.)
Note that these are all grouping constructs. A single entity is not a communicative device. As soon as there are two (or more) things the relationship between them has to be shown. (I am happy to be argued out of this one, but then I want some decent examples of singular CDs.)
In addition to these examples of CDs, the most important information we can provide is how the different CDs can be combined together. For example, a title CD is subtly different from a heading CD. Heading CDs can be children of many other CDs (to be specified) but a title is the "start" of something. Perhaps better said, if you are creating a CD tree (CDT) from scratch, then the first element has to be a title or scene element, and is not allowed to be a label element.
Some questions. Are title and heading really different? Are menu and heading really different? Do we want ordered and unordered scenes or variations of a single scene (same sort of discussion as "element or attribute" in XML)?
Having, I hope, conveyed an idea of what these may look like and how they fit into the grander scheme of things, now we need to figure out the requirements. This is an initial list.
A particular communicative device can be used for expressing various discourse structures. A particular discourse structure may have a preferred communicative device, but could use one of several. In order to fully express a discourse structure, perhaps several comm devs need to be combined together.
To start with, we need to be able to build up a tree of these things. For each CD we thus need to know either which children it is allowed to have, or what its parents may be. I leave it to the architects among us to tell me which the better method is.
Attention overload. When deciding how to combine and nest the CDs we need to bear in mind that any media that we use don't clash with the rules for overloading the human reader. This extra addition to the constraint layer was thrown out of the HFOT, but it is certainly something that cannot be evaluated in the CDT until we know the spatial/temporal presentation of the media items.
Graphic design. This needs to be woven into the picture. For example, a title CD will have a very clear transition (e.g. fade in from black), applying to the title and the content, and probably overriding the transitions specified further down the CDT. Each CD will have a list of potential transitions supplied with it. However, even these should not be specified at the "fade-to-black" level, but at a CD level, such as "major transition" or "minor transition". It is up to the graphic design style being used at the time to specify the transitions which can be used in the presentation. This implies there is a separate style module (which doesn't surprise me).
Given a CDT we need to know how to divide it up into the HFO level. For example, the same CDT may fit into 3 PC screens or 10 PDA screens. Perhaps lots of constraint-solving at the HFO level is precisely the solution to our problem.
Do not underestimate the concept "importance" when dealing with the build up of the CDs. I haven't included any graphic design references here yet, but when figuring out how to display information the relative "importance" of items was very significant. It allows relative font sizes and spatial layout to be chosen. Perhaps the whole CDT will boil down to a tree of "this is more important than that", and the rest is just left up to the constraint-solving.
My proposed next step is to sniff around some MM presentations and some web-sites and see if my example CDs are at all realistic. I also want to build up a more concrete single CD to check the connections with the graphic design side.
After a long discussion with Jacco we came up with generalizing the communicative device as follows:
It is up to some other mapping process to convert the importance and ordering into some space/time/link specification. This seems suspiciously like the three things in Smart Style: layout (space/time/link), content (the children?) and presentation structure (grouping=composition)?