Explores the use of linear models and a combination of neural networks and linear classifiers to create a hybrid hierarchical mixture of experts (HME) model. Results confirm that using the hierarchical structure of the classification vocabulary improves categorization performance. (AEF)

Examines historical and current difficulties in information retrieval, including the tendency of technology, because of the problems of scale and distribution, to make information invisible to users. Attempts to organize networked resources are described, and serendipity is proposed as a model for future retrieval. (two references) (MES)

This procedures manual provides a list of relevant reference tools, as well as procedures and criteria for (1) document acquisition and assignment, (2) descriptor and identifier control files, and (3) program and human resources files. The appendices include examples of a document resume, a program resume, and a human resources resume. (JEG)

The Broad System of Ordering (BSO), a subject-indication coding and ordering scheme developed to meet the requirements of the UNISIST program for an international switching mechanism between information systems using diverse indexing/retrieval languages is described. The scope and purpose of the BSO, its development and testing, a schematic…

The objectives of the seminar were to report on: (1) the American Institute of Physics (AIP) Universal Decimal Classification (UDC) project work; (2) the UDC work, trends and opinions in Germany, the United Kingdom and other countries and (3) the production and development of UDC schedules and indexes, and Perreault's work on "relators." Twelve of…

Presents the results of an investigation of the use of links on Web pages; a content analysis was performed on 75 Web pages and their links. Results will provide a conceptual framework in which Web page design can be considered, including issues of authoring, retrieval, and built-in and on-the-fly guidance for reading and browsing. (Author/AEF)

Examines usage scenarios for a database of descriptive information to access networked information resources such as online library catalogs, file archives, online journal article repositories, and information servers. Existing classification schemes and their usefulness for networked information resources are discussed. Specific data elements…

Characterizes information retrieval systems in a network environment from a user's perspective along two axes: query and presentation. Several examples of early, current, and developing information retrieval systems are considered within this framework, and Daltext, a prototype system to support research on text-based retrieval systems, is…

Cooperation in Documentation and Communication (CODOC), a computerized system designed to provide a quick and economical method of in-depth access to individual publications within a government documents collection, currently operates as an information network in Ontario and Quebec and independently in several other libraries in Canada and the…

The stress of this symposium was on the role of the Universal Decimal Classification (UDC) and other indexing languages, especially in multilingual and multidisciplinary information systems, rather than on its potential uses in mechanized retrieval systems. The consensus of the Symposium was: (1) that a system like the UDC cannot, without major…

The four chapters included in this report are based on LINCS project activities undertaken since 1968, with an emphasis on indexing tools in the language sciences and related problems. Chapter one, "Indexing Tools for the Language Sciences: Methodology," discusses the development of a LINCS thesaurus and its role in the LINCS network.…

Presents abstracts of SIG Sessions. Highlights include digital collections; information retrieval methods; public interest/fair use; classification and indexing; electronic publication; funding; globalization; information technology projects; interface design; networking in developing countries; metadata; multilingual databases; networked…

Describes plans for an investigation of gopher protocol use for browsing and searching for information available via the Internet. Use of classification schemes based on traditional library models will be compared to the use of an existing gopher. Research questions, literature on related issues, construction of the classification schemes, and the…

Argues that library skills and methods used for organization of information are needed to organize Internet information resources. Four aspects of information organization are discussed: recording existence and identity of information bearing entities; acquiring these entities; providing name, title, subject, and other access information; and…

A report is presented of a study undertaken to analyze the future clearinghouse requirements of the Educational Resources Information Center (ERIC) system. Volume 1 describes research on the domains of future ERIC clearinghouses. A taxonomy of education was created, a sample of educational literature categorized by means of it, and 21 domains…