Augmenting the analyst via situation-dependent reasoning with trust-annotated facts

We say that a computer program augments the analyst if it can infer facts that are implicit in existing information, but that may be relatively difficult for a human to infer. Among a multitude of reasons, the analyst´s task is difficult because (1) reported information to be analyzed and reasoned about often cannot be completely trusted (requiring verification attempts via further collection of information, corroboration where verification is not possible, and/or assumption-based reasoning) and (2) evaluation of trust (reliability, credibility) is context (or situation) dependent. These two sources of difficulty, and the possibility of augmentation-generated false alarms, imply that if one wants to employ the capabilities of an automatic reasoner, the reasoner must be able to deal with these kinds of complexities.