Collaborative Project: Managing Authority Lists for Customized Linking and Visualization

Johns Hopkins University, Baltimore MD

Investigators

Abstract

This is a collaborative project with Tufts University (Award No. 0226304). This project is developing Services for a Customizable Authority Linking Environment (SCALE) to support all levels of reading in the National STEM Education Digital Library (NSDL). In particular, the project is developing two broad classes of service. First, the investigators are providing automatic linking services that automatically bind key words and phrases to supplementary information. Such services help students, professionals outside a particular discipline, and the interested public to read documents full of unfamiliar technical terms and concepts. For example, astronomy students and curious amateurs may need to see expansions of acronyms--e.g., MACHO (massive compact halo object)--or pictures of "Kuiper belt objects"; or the student struggling with research papers on bioluminescence may need to locate information about particular chemical processes or relevant species of echinoderms. Second, the investigators are basing automatic linking on authority control of names and terms and on links among different authority lists such as thesauri, glossaries, encyclopedias, subject hierarchies, and object catalogues. Activities in the project include: * creating and maintaining authority lists of technical terms and concepts harvested from Open Archives Initiative (OAI) metadata, extracted from full text, and imported from existing authority lists; * extending current automatic hypertext capabilities to embed glosses of technical terms and links to related passages within HTML, PDF, and XML documents in the NSDL; * providing term detection and document linking through Simple Object Access Protocol (SOAP)-based Web services; * evaluating the interface to and functionality of SCALE in cooperation with the National Virtual Observatory; * customizing SCALE through explicit user configuration and adaptive learning of preferences; and * collecting annotations on the quality of links to improve precision and provide a training set for future systems. Much of the work at the Tufts University Perseus Digital Library Project (http://www.perseus.tufts.edu/) and the Johns Hopkins Digital Knowledge Center (http://dkc.mse.jhu.edu/) has already focused on exploiting various kinds of authority lists (gazetteers, biographical dictionaries, dictionaries, glossaries of technical terms, and name authority files) for the automatic generation of hypertext links and for visualizations such as automatically generated dynamic maps and timelines. Such link generation complements the current practice of automatic identification and aggregation of citations. The current project augments and transfers the existing technology for managing authority lists, converting this from a research effort to an institutionalized service serving a wider community.