Map Specialization Step 1: Design the Map Element Types
A typical FAQ consists of one or more groups of questions, where each group has a descriptive title. There may also be some introductory material that is not itself FAQ questions.
That suggests that an FAQ map needs the following components:
- A topicref for referring to individual FAQ questions, e.g. "faq-question"
- A topic head type for creating sets of related questions, e.g. "faq-question-set"
- Some metadata for the FAQ set as a whole
- The ability to refer to generic (non-FAQ question) topics at the start of an FAQ set.
The metadata requirement is handled by the generic metadata for topicrefs and maps
(at least for the purpose of this exercise). References to generic topics can be done
using unspecialized <topicref>
.
That just leaves two element types to be defined for the FAQ map domain:
<faq-question>
- Represents the use of a single FAQ question topic. Would normally be used with
faq-question topics. Because we've defined the
<faq-question>
topic type to require short descriptions, it is probably appropriate to set the default value for the@type
attribute to "faq-question" so that users are warned if they point to other topic types. Because FAQ questions are explicitly atomic,<faq-question>
does not allow subordinate topicrefs. <faq-question-set>
- Creates a group of related FAQ questions. May have an initial generic topic or may
refer
to a non-FAQ-question topic to act as an introduction or to satisfy the need
for title-only topics in some processing environments.
The content model is:
(topicmeta?, topicref?, faq-question*)
.
For the FAQ map type we need the following element types:
<faq-map>
- Specializes from
<map>
and serves as the root of an FAQ publication. It's content model is:1(title, topicmeta?, keydef*, faq-question-set+, reltable*)
1 I like to define my map types so that they allow
exactly one root topicref type that serves as the root of the navigation
tree defined by the map. I think that makes the use and intent of the
map clearer, especially now that the
@processing-role
attribute makes a clear distinction between topicrefs that contribute to
the navigation tree and those that do not (resource-only topicrefs). It
also makes it clearer what the effect of a mapref to a subordinate map
is because this pattern means that a reference to a map is equivalent to
a reference to the root navigation topicref in the map. However, I did
not apply that pattern to this example in order to keep it as simple as
possible. For example, in a real design I would probably make the
content model of <faq-map>
something like "(faq
| faq-question-set)" where <faq>
is
"(faq-question-set)+". This would let you have a
single root map with <faq>
as the root of the entire
navigation tree but also allow submaps that contain single
<faq-question-set>
elements. That model allows a
document with multiple FAQ question sets, but ensures that all maps have
exactly one navigation root
topicref.