This effort will apply an innovative context-sensitive spreading-activation measure to compute relevance measures to rank information objects. It will also use psychometrically measured preference functions to update both query characterizations and information object characterizations. Phase I will demonstrate the ability of spreading-activation to rank objects available for retrieval and reflect psychometrically measured preference functions of the query submitted on a UNIX workstation. Phase II will extend them to wide-area information server (WAIS) architectures using the ANSI Z-39.50 protocol and to multimedia objects including audio, video, photographs or any other data relevant to a stated request. The resulting information retrieval system will be capable of extending mental models NASA scientists and engineers have of their information needs to distributed heterogeneous data in WAIS databases and other data sources, such as those associated with the CERN World Wide Web project. This will include usage-based dynamic updating of the weighted association network used to find and rank information available for retrieval.Include Memory Extension (ME) systems (applications) for UNIX, MS-Windows, and Macintosh work-stations connected to networks, such as the Internet. Daemons that would reside on remote systems to support spreading-activation searches and associated characterization network link weight updates. Value-added products for improving searches by users of commercial information services, such as Dialog and CompuServe, are also possibilities, along with more effective library search systems.Information Retrieval Spreading Activitation Memory Context Ranked SoftwarePhase 2 conversion
