WP 2 - The AMIDST framework for representation and reasoning
- Develop a general probabilistic model class (the AMIDST framework) that can support efficient methods for doing reasoning in data streams.
- Ensure that the resulting framework can be instantiated into models that capture the particular aspects and properties characterizing the three use cases.
- Implement the developed modeling framework in the AMIDST toolbox.
The work in this work package will be based on the data characteristics and requirements analysis established in Work package 1 for each of the use case providers. Specifically, we will develop a general probabilistic model class that can support the use cases and encompass static as well as dynamic models having both bounded and unbounded time-horizons. The framework will be implemented in the AMIDST open source toolbox and designed to support an easy integration with the inference and learning algorithms that will be developed in Work package 3 and 4.