Efficient Reasoning with Ambient Trees for Space Exploration

Abstract

Modern reasoning is based on inference techniques such as induction, deduction, abduction, subsumption, classification and recognition. These inference techniques are very inefficient when applied to large amounts of knowledge such as ones employed by contemporary unmanned spacecraft. For efficient reasoning, we aim at knowledge representation based on special ambient trees determining special knowledge contexts to help such spacecraft retrieve context-relevant knowledge and perform deductive reasoning, which would not be otherwise highlighted. Contexts via their ambient trees provide a sort of a condensed and explicit symbolic representation of the world. This representation is cleaned from the overwhelming information that is non-relevant to the context and thus, it provides for efficient models of situations to reason about.