Bert R. Boyce

1163

Special Topic Issue: Integrating Multiple Overlapping Metadata Standards
Guest Editor: Zorana Ercegovac

Introduction
Zorana Ercegovac

Metadata has remained one of the critical components in the context of knowledge representation and data mining in digital libraries as it had traditionally been in the context of the pre-Web libraries. Today in the digital libraries environment in which individual collections of massive heterogenous objects need to be unified and linked in a single resource, we have witnessed both the growth of different metadata and the attempts to reconcile the common attributes in the existing overlapping standards. The ultimate goal is to make it possible to access relevant information seamlessly regardless of its type (e.g., visual and museum objects, historical data, cultural heritage, scientific data), location, and scholarly tradition (e.g., librarians, archivists, scientists).

This Special Issue of JASIS addresses different applications of metadata standards in geospatial collections, education, historical costume collection, data management, and information retrieval, and explores the future thinking of metadata standards for digital libraries.
 

1165

Collection Metadata Solutions for Digital Library Applications
Linda L. Hill, Greg Janee, Ron Dolin, James Frew, and Mary Larsgaard

We begin with an article by Hall and her colleagues at the Alexandria Digital Library. They first look at the meaning of the concept of a collection in the context of digital libraries in general, and especially within the Alexandria Digital Library. In order to make sense of high heterogeneity that exists among digital library collections, Hall et al. discuss the design and implementation approach of collection metadata that "represents the inherent and contextual characteristics of a collection." Since ADL's collection contains maps, remote- sensing images, aerial photographs, and related texts, the architecture of the ADL collection metadata differs from either the archival/EAD approach or the START's text-oriented approach. The article defines its structure and explicates ADL's capability to support collection registration, network discovery, user documentation, and collection management for its georeferenced collections.
 

1169

Conceptual Design and Deployment of a Metadata Framework for Educational Resources on the Internet
Stuart A. Sutton

In order to link teachers and educational material that is distributed across the Internet and created by federal, state, academic, nonprofit and commercial sites, Sutton's paper discusses a conceptual design and employment of a metadata framework for educational resources on the Internet. The paper first describes the Gateway to Educational Materials, GEM, framework (http://geminfo.org/) and its underpinnings in the Dublin Core Set; the paper goes on to suggest the extension of the original GEM to account for the information seeking behavior of teachers who search for educational resources on the Internet. The result was the proposal to add unique metadata elements: five description elements, two evaluative elements, and one meta- metadata element. Sutton also discusses syntax, design and implementation of harvesting tools for retrieving GEM metadata.
 

1182

Metadata Elements for Object Description and Representation: A Case Report from a Digitized Historical Fashion Collection Project
Marcia Lei Zeng

Zeng examines the fitness of three existing metadata formats (USMARC, The Dublin Core Element Set, and the Visual Resources Association) to support a collection of historical fashion objects held at the Kent State University Fashion Museum. Zeng adopted and modified the VRA metadata format to catalog the entire digitized historical fashion collection.
 

1193

A Comparison of the Two Traditions of Metadata Development
Kathleen Burnett, Kwong Bor Ng, and Soyeon Park

Burnett, Ng and Park discuss two different approaches: the bibliographic approach that has origins in cataloging (the library community), and the data management approach that has roots in computer processing (the computer science profession). The article compares element sets between and among six different metadata (i.e., USMARC, The Dublin Core, TEI, Semantic Header, IAFA Templates and URC) and supports a proposal for an integrated approach to metadata.
 

1209

Use of Metadata Vocabularies in Data Retrieval
Edwin M. Cortez

Finally, in the context of Information Retrieval and the Internet, Cortez considers a metadata vocabulary as a negotiator between a set of 39 different databases (disparate by structure, vocabulary, use and purpose) and equally diverse user populations. The proposed metadata vocabulary relates to the domain of food, agriculture, natural resources and rural development; it attempts to normalize semantic and hierarchical distinctions between and among different databases and to act as a front-end unified language to the prototype Database Catalog.

1218

The Ecological Approach to Text Visualization
James A. Wise

The Spatial Paradigm for Information Retrieval and Exploration, SPIRE, converts digitized text documents into vector space document representations using 280 element vectors whose elements were produced by a neural net trained on the domain of the documents. These are clustered with a similarity measure and projected onto a two-dimensional plane using a modification of multidimensional scaling that uses document-to-centroid distances rather than pairwise document distances. The visualization shows the reoccurrence of a concept as a height on a projection that resembles a terrain map.

A Hybrid Method for Abstracting Newspaper Articles
James Liu, Yan Wu, and Lina Zhou

Liu, Zhou, and Wu begin their abstract extraction from Chinese text by comparing character pairs with user chosen keywords for exact, partial, or variable character matches. Word frequency of all words is compared to a standard word frequency table, where nouns and verbs of frequency at variance with the standard are extracted . High variance words are used to select sentences until the required length of text is extracted. In a combined method, matching is supplemented by weighted extraction. An additional level uses parts of speech, pronoun referents, and syntactic rules as well as syntactic markers explicit in Chinese text.

Thirty five users were surveyed and 60% found keyword and percentage extraction to be useful. The extraction of summaries was not well received.
 

Formal Features of Cyberspace: Relationships between Web Page Complexity and Site Traffic
Erik P. Bucy, Annie Lang, Robert F. Potter, and Maria Elizabeth Grabe

Using a sample of 5,000 Web sites top ranked by hits using 100hot's InSite Pro service, Bucy, Lang, Potter, and Grabe randomly selected 500 sites, and 496 home pages were coded to reflect domain name, rank, average number of page views over six weeks, and the banner, body, and advertisements were analyzed for features and links. Banners occur on 75% of sites and are most commonly white. One-fifth featured movement. Home pages averaged 2.4 screens in length and 79% used one or more frames. The dominant background color was white. A graphical element occurred on 95% of the pages, with a logo being the most common. Movement was present in about one-third of the pages. Asynchronous elements--links, surveys, contact information--occurred in 98.9% of pages with an average of 27.1 such elements per page. Just 15.9% used real-time interactive elements, like audio or video links, or chat rooms (which were the most common of these). Over half the pages exhibited advertisements of some kind but less thanone-third of these had dynamic features.

For commercial sites, high visitation correlates with high graphics use and less strongly with asynchronous interactive elements. In noncommercial sites, there is a strong correlation between visits and asynchronous interactive elements. Real time interactive elements are rare. Advertising is prominent, but pages are not generally over-designed.
 

1224

1234

1246

Understanding Information Retrieval Interactions: Theoretical and Practical
Implications
Carol A. Hert
 

Information Literacy: Essential Skills for the Information Age
Kathleen L. Spitzer with Michael B. Eisenberg and Carrie A. Lowe
 

Scholarly Book Reviewing in the Social Sciences and Humanities. The Flow of Ideas Within and Among Disciplines
Ylva Lindholm-Romantschuk
 

1257

1257

1259

© 1999 , American Society for Information Science
Last update: September 23, 1999