Parsimonious downgrading and decision trees applied to the inference problem
LiWu Chang, Ira S. Moskowitz · 1998
In this paper we present our new paradigm for dealing with the inference problem which arises from downgrading. Our new paradigm has two main parts: the application of decision tree analysis to the inference problem, and the concept of parsimonious downgrading. We also include a new thermodynamically motivated way of dealing with the deduction of inference rules from partial data. Keywords Data mining, inference, downgrading, rules. 1. A NEW PARADIGM Our new paradigm is a combination of decision tree analysis and parsimonious downgrading. Decision tree analysis has existed in the field of AI since the 1980's [3]. In brief, decision trees are graphs associated to data, with the goal of deducing rules from the data. Our new paradigm applies decision trees to the inference problem. In this paper we introduce the new concept of parsimonious downgrading. When High wishes to downgrade a set of data to Low, it may be necessary, because of inference channels, to trim the set. Parsimonious...