Suppose you want to create a data model about restaurants. One approach is to create a list of questions that people typically want answered when searching for a restaurant:
1. Where is the restaurant?
2. What type of food do they serve?
3. How much does it cost?
4. When are they open?
5. How to contact them?
A direct mapping from question-to-XML yields:
<Restaurants>
<Restaurant>
<Where-is-the-restaurant>...</Where-is-the-restaurant>
<What-type-of-food-do-they-serve>...</What-type-of-food-do-they-serve>
<How-much-does-it-cost>...</How-much-does-it-cost>
<When-are-they-open>...</When-are-they-open>
<How-to-contact-them>...</How-to-contact-them>
</Restaurant>
...
</Restaurants>
Is that a reasonable approach to creating some data models - create a list of questions that users typically want answered and then directly map the questions to element names?
I think that, this would be a great approach, to create XML data models for many domains. Happy to learn this from you.
Advantages/disadvantages?
Generally, I think that at the beginning of implementing an XML based solution (or for that matter, any software application solution), developers struggle a bit to have a reasonable approach to create data model for a business IT problem. The advantage of your approach is, that it is simple and would work for many domains.
Your approach I think, is not enough, to discover the final data model for software applications. The final data model, might require for example, knowing certain dependencies between sub domains, which your question based approach might not be complete in itself.
