Mechanisms for associating metadata with network resources, and for locating the network resources in a language-independent manner, are disclosed. Owners of network resources define metadata that describes each network resource. The metadata may include a natural language name of the network resource, its location, its language, its region or intended audience, and other descriptive information. The owners register the metadata in a registry. A copy of the metadata is stored on a server associated with a group of the network resources. A copy of the metadata is stored in a registry that is indexed at a central location. A crawler service periodically updates the registry by polling the information on each server associated with registered metadata. To locate a selected network resource, a client provides the name of the network resource to a resolver process. The resolver process provides to the client the network resource location corresponding to the network resource name. Multiple metadata mappings can be established for the same network resource, in which each mapping stores a name expressed in a different natural language. Accordingly, network resources can be located merely by providing the name of the network resource in any natural language that is convenient for the client.