Building Hierarchical Classifiers Using Class Proximity

In this paper, we address the need to automatically classify text documents into topic hierarchies like those in ACM Digital Library and Yahoo!. The existing local approach constructs a classi er at each split of the topic hierarchy. However, the local approach does not address the closeness of classi cation in hierarchical classi cation where the concern often is how close a classi cation is, rather than simply correct or wrong. Also, the local approach puts its bet on classi cation at higher levels where the classi cation structure often diminishes. To address these issues, we propose the notion of class proximity and cast the hierarchical classi cation as a at classi cation with the class proximity modeling the closeness of classes. Our approach is global in that it constructs a single classi er based on the global informationabout all classes and class proximity. We leverage generalized association rules as the rule/feature space to address several other issues in hierarchical classi cation.