Techniques to assign a ranking value to objects in a database such as a collection of cross referencing documents, the World-Wide Web or a hyperlinked database are described. The ranking value assigned to a given data object represents a cross-cluster strength metric and is a function of the object's importance across all groups or clusters in which the object is classified. The cross-cluster strength metric may be particularly beneficial in enhancing the performance of web-based search engines because it emphasizes the importance of objects that appear in multiple groups while de-emphasizing the importance of objects that, while highly linked within one or a few groups, are relatively unlinked to objects in other groups.