The natural environment provides an enduring context for our everyday lives. Whether we build some product that is designed to endure whatever nature has to offer, or we have to endure it ourselves, knowledge is our best bet to maintain the upper hand. In the lab, we always used as our battle cry “Nature’s fighting us, we must be close!”. Well, that is what this place is about.
Context Modeling for Land, Aquatic, and Atmospheric Environments
The research featured the following novel findings and approaches
- Spectral decomposition approach to model semi-Markov terrain features.
- Terrain elevation correlation approach applied to geospatial regions.
- Pattern-based stochastic approach to model sample spaces.
- Use of maximum entropy to model distributions with very concise formulations.
- Depth-corrected wave height model applied to various bodies of water.
- Diffusion model of oxidation and corrosion applying uncertainty quantification.
- Semantic organization of models using SWEET ontology.
- Semantic web server for model search and interaction.
- A logic-based domain specific language for formal specific and model generation.
- Phrase-based matching of requirements to context models.
We find that applying an ontological classification strategy to stochastic models of the environment provides a simple, efficient while comprehensive approach while a workflow-driven semantic web provides services.
I want to see how far we can stretch concepts such as the Semantic Web, a knowledge network that is more organized than the ad hoc network than we are currently dealing with. The rationale for the concept of the Semantic Web is to categorize and classify knowledge so as to minimize misinterpretation and ambiguity among concepts. Earth sciences is a great place to start, as it includes a wide range of topics, but also contains a lot of conflicting knowledge.
Context/Earth is essentially a Semantic Web project, where we organize interactive environmental models (of land,ocean,atmosphere) according to an ontology. We worked with the people at NASA JPL, and incorporated their SWEET ontology (Semantic Web for Earth and Environmental Technology) into a semantic web server which we call the Dynamic Context Server (DCS).
One instance of the DCS hosted on a cloud server here:
Cloud DCS instance
The name derives from the fact that a context model is a model of the surrounding environment which one can use to develop systems or analyze data.
This is all open-sourced code and we use this blog to keep a running commentary on how the project is progressing now that the original development phase ended.
The bottom-line is that this kind of knowledge repository is going to transform the way that we use and apply environmental and earth science information in the future. Climate science is not the only game in town when it comes to the environment. The earth is a dynamic system and the natural resources and fossil fuel reserves also play a big role, and understanding how we can leverage what the earth provides is a good path to follow.
The question that I pose is: Can this knowledge be deconflicted and disambiguated so that the truth can emerge?
The hope is that semantic knowledge will help resolve the uncertainty and ambiguity from various sources of information.