In a distributed computer system, an automated document annotation system and method adds hypertext cross-references to a set of known information sources into documents requested by a client computer in such a way that the merged document is displayable by existing Web browsers. The distributed computer network incorporates a plurality of servers to store documents. Each stored document has a unique document identifier and is viewable from a client computer having a browser configured to request and receive documents over the network. An annotation proxy, which is a software procedure configured to merge a requested document from a first server with hypertext links to documents containing associated supplemental information. The set of hypertext links and criteria for identifying where such links should be added to requested documents are defined by one or more dictionaries of cross-references. The annotation proxy then relays the merged document to a receiver unit that is selected from another proxy, such as a firewall proxy or another annotation overlay proxy, or the browser, which ultimately displays the merged document. The annotation proxy optionally includes a dictionary generator that generates a dictionary of references to documents requested by the user, each reference in the dictionary indicating the textual context of the hypertext link or links used to request the associated document. The generated dictionary represents information sources known and used by the user. The annotation proxy then annotates requested documents with cross-references in the dictionary that was generated by the annotation proxy.