Predicting the Basic Level in a Hierarchy of Concepts

The “basic level”, according to experiments in cognitive psychology, is the level of abstraction a hierarchy concepts at which humans perform tasks quicker and with greater accuracy than other levels. We argue that applications use concept hierarchies could improve their user interfaces if they ‘knew’ are basic concepts. This paper examines what extent can be learned from data. test utility three types features, were inspired by theory: lexical structural features frequency features. evaluate our approach on WordNet, create training set manually labelled examples different part WordNet. Our findings include accurately identified within one domain. Concepts difficult label for also harder classify automatically. provide insight into how classification performance across parts improved, necessary identification larger scale.