Predicate Logic-based Assembly of Situation-specific

Applications such as decision support systems and data fusion must reason about inherently uncertain properties of entities in a domain of application, as well as their behavior and relations. Tasks such as information fusion for battlespace awareness require both reasoning under uncertainty and logical reasoning about discrete entities. Probability is the most well-understood and widely applied logic for computational scientific reasoning under uncertainty. However, appropriate application of probability theory often requires logical reasoning about which variables to include and what the appropriate probabilities are, in order to maintain tractably-sized models. To address this need, Information Extraction and Transport, Inc. proposes a Predicate Logic-based Assembly of Situation-specific probabilistic Models for battlespace C2 Assistance (PLASMA) architecture for combining logical and probabilistic reasoning with discrete and continuous variables to perform data fusion and decision support by supporting parsimonious situation-specific probabilistic model construction guided by first-order logic model construction specifications. The architecture applies logical reasoning to identify relevant variables for the constructed network, and to build the Bayesian network from a knowledge base of Multi-Entity Bayesian Network fragments, and constructs situation-specific models using probabilistic generalizations of forward-chaining and backward-chaining of rules, encoded in first-order logic suggestors.