IET developed Quiddity*Modeler, a probabilistic frame-based representation language, for use in dynamic situation modeling tasks. Quiddity*Modeler is a Bayesian decision network software package that implements SPI algorithms in JAVA using an object oriented language framework to easily assemble or dynamically modify a Bayesian network or a decision network. Specifically, Quiddity*Modeler features include:

• Supplying capabilities for specifying knowledge models with uncertainty, using a frame-based language
  

• Supporting representations of hypotheses (e.g, association, type, existence)
  

• Supporting compositional model generation
  

• Constructing Bayesian networks from frame instances to answer queries
  

• Providing three major facilities: (1) a domain modeling language; (2) a set of situation modeling facilities; (3) a facility for querying situation models
  

• Expressing domain knowledge as fragments of Bayesian networks in a modular and compact way, facilitating reuse
  

Instances of probabilistic frames are created dynamically for each situation, allowing situation specific probabilistic inference. The probabilistic inference is performed by Quiddity*Inference using a Bayesian network constructed dynamically from the current set of probabilistic frame instances. The generation of Bayesian networks utilizes Quiddity*Inference’s local expressions to exploit independence relationships to speed up the inference. Quiddity*Modeler is also fully integrated with Quiddity*Script, allowing the user to define frames, create instances, and make situation-specific queries interactively.

Back to Tools Page