Abstract: Large stores of information are often organized in a hierarchical taxonomy to aid a search and retrieval of the information. The hierarchical taxonomy generally consists of related categories of information, called “nodes,” that each may contain information relevant to the search. Each node is addressable according to its path in the hierarchical taxonomy. In information stores where the number of nodes having relevant information is extremely large, such as the Internet, providing a cohesive, intelligent, and organized display of the search results becomes extremely important to the success of a user traversing the store to find relevant information. The invention provides such search results by ranking each node of the taxonomy to determine which nodes are most likely to be relevant to the search request. The invention then creates a conceptually-related “cluster” of nodes by selecting a relevant “seed” node and relevant nodes related to the “seed” node.

Abstract: A system reads a text string from an address entry location of a web browser and determines whether the text string identifies a web page. If the text string identifies a web page, then the identified web page is retrieved. If the text string does not identify a web page, then the system identifies multiple web pages associated with the text string and identifies a best web page from the multiple web pages associated with the text string. The best web page is displayed in a first pane and at least a portion of the addresses associated with the multiple web pages is displayed in a second pane. The system also determines whether the text string to identifies a resource on a local network. An attempt is made to modify the text string to identify a web page if the text string does not already identify a web page. The best web page can selected from a web directory search, a web page search, or a combination of both.