There are different ways to organize bottom-up markup. A technical solution is namespaces, but that just keeps the chaos from bugging the bots.
Namespaces! A linguistic solution without behavioral context (e.g., requirements). So many different implementation models. So much fun. | ||
| --Anonymous source | ||
Another approach is negotiations. As individual markup is shared and encounters similar/ overlapping markup, conversations are used to generalize the concepts and markup. Different concepts get different names. Individual markup feeds departmental standards. Semantic generalization flows up through layered DTDs, matching the organizational hierarchy. This represents an evolutionary approach to bottom-up data modeling and system design.
Coming at the question from the opposite direction, individual markup could be anchored to a separate, semantic authority. At the first Ontolog face-to-face meeting & workshop in 2003, Adam Pease introduced SUMO, the Suggested Upper Merged Ontology to the group. The aftermath is described in the forward to In Adam's book Ontology:
The result of [Adam's] behavior was quite infectious. Adam, Kurt Conrad, and I ended up in a late night sushi restaurant somewhere near Menlo Park, CA, discussing how to map SUMO concepts to Mandarin, Japanese, and Cantonese, how WordNet can reference SUMO, and why First-Order Logic (FOL) constraints are generally a cool concept to have in advanced computer systems. An upper-level ontology, such as SUMO, is a common, shared conceptualization of a domain. | ||
| --Duane Nickull | ||