Abstract for: Structured Semantic Descriptions of Systems

Our goal is to develop structured semantic descriptions of dynamic systems. Such descriptions would be useful for developing portable, extensible, and reusable models as well as for supporting user interaction. These models build on rich ontologies that include rules to support state changes. Thus, they are related to object-oriented modeling from software engineering. Moreover, our models go beyond semantics to include discourse tags and could be used to provide explanations and tutorial presentations adapted to a user’s background. Here, we examine developing highly structured models for natural physical systems. Specifically, we are developing a rich semantic model using Python for the Snowball Earth theory. Our model describes processes and causal flows for a geological era when it is hypothesized that the earth was completely covered by ice. Rather than developing a high-resolution qualitative model, we aim to implement a quantitative model. We believe it will extend our previous work on highly structured alternatives to traditional text research reports. Potentially, this work will also provide a rigorous alternative to statistical AI text generation approaches.