The virtual object model for distributed hypertext

[摘要] Hypertext systems have traditionally been characterized as collections of data connected by ;;links,;; or navigational paths between explicit anchors in the data; links from hypertext systems to other resources or to computation engines have been afterthoughts.The thesis of this dissertation is that a model that views hypertext ;;links;; as queries can encompass not only the traditional definition of hypertext, but also integrated access to remote resources, as well as computation and intelligent retrieval of data from within and without the hypertext system. The Virtual Object Model is presented in support of this thesis. The Virtual Object Model provides a comprehensive model that includes naming and typing of data in local and remote information sources and computations, thus facilitating both human and automatic search and retrieval in hypertext.An implementation of the model, called ;;Amanuensis,;; includes query engines that support query of diverse remote resources including MEDLINE (TexSearch), the Unified Medical Language System (UMLS) Metathesaurus, and SNOMED nomenclature, in addition to two distributed hypertext systems, the Virtual Notebook System (VNS), and World-Wide Web. Interfaces to Amanuensis were created from the VNS and Mosaic a Web viewer. The utility of Amanuensis has been demonstrated by the development of the prototype of a Distributed Medical Dictionary that provides for intelligent construction of medical terms from a hierarchy of vocabulary resources including the UMLS Metathesaurus.Experience with Amanuensis demonstrates that the Virtual Object Model simplifies naming and searching across diverse domains. It simplifies the user interface because objects from diverse sources can share similar behaviors. This can encourage the use of network bandwidth in place of local storage, reducing users' personal or institutional needs for disk space and lessening the individual burdens of data management. Searching in a large, diverse information space is still inherently difficult, but the Virtual Object Model integrates the information space better than previous hypertext models, simplifying the task of constructing intelligent agents that aid the user in information gathering.

[发布机构] Rice University

[效力级别] science