1

I'm writing game worlds and I've started working on representing the worlds not just as text and images but as a graph of topics and associations. In other words, an ontology representing the game world's characters, places, events, concepts, terms and so on.

Where I've got a bit stuck is in defining and naming the relationships between topics. It's easy enough to come up with things like "is a", "part of", "located in" etc, but as the work goes on, I realize that using the terms loosely will not work well, there are many relationships that overlap in meaning and you start wondering if this hasn't already been done. I've looked into OWL for creating ontologies, and topic maps, but what I lack is an actual data set of named associations (predicates in RDF) that I can build on, that have been vetted and used for larger projects.

What is a good strategy and resource to describe relationships between concepts in an ontology?

RipperDoc
  • 694
  • 8
  • 16

1 Answers1

0

Schema.org is probably the definitive resource when it comes to widely accepted nomenclature for ontologies. See their page for Organizations for a good example of their property names.

I found a similar question that was addressed to the Neo4j/Cypher graph database audience, but may provide some good insights as well. A key one is the trade-offs between fine-grained relationships ('LIKES_POST') vs course-grained relationships ('LIKES'). The finer the grain, the greater the specificity, but the greater the overall complexity of the graph. In general, these trade-offs depend on your use case, where you can determine which direction best suits your needs.

Chilangosta
  • 101
  • 8