SGER: Concise, Intelligble, and Approximate Modeling of Large Classifications in Natural and Social Science

点击翻译按钮获取中文摘要

In this exploratory research, an attempt is being made to understand the generic scientific task of classification. This task is perhaps somewhat better understood than discovery, which has previously been studied by the same PI. In the AI field, much research is going on in induction of classifications and in classification learning. But it is important to augment what is discovered in these areas by looking at two other important aspects of classification: describing classifications (i.e., developing a good model of the classification scheme), and relating a classification to new criteria (e.g., relating a biological classification based on characteristics or serology to one based on DNA). This work will lead to a better understanding of the cognitive aspects of classification, as well as to new computational tools for aspects of classification that can be used by scientists in a variety of disciplines.