Explains how a competition-based connectionist model for diagnostic problem-solving is adapted to information retrieval. Topics include probabilistic causal networks; Bayesian networks; the neural network model; empirical studies of test collections that evaluated retrieval performance; precision results; and the use of a thesaurus to provide…

Describes a session on information retrieval systems that planned to discuss relevance measures with Web-based information retrieval; retrieval system performance and evaluation; probabilistic independence of index terms; vector-based models; metalanguages and digital objects; how users assess the reliability, timeliness and bias of information;…

Reviews the development of research and evaluation in information retrieval. Highlights include the Cranfield projects concerning recall and precision, the Medlars experiment that focused on relevance, TREC (Text Retrieval Conference), evaluation versus understanding, interface versus functionality, document length, and terms and user interaction.…

Discusses an experiment with the "Thesaurus of ERIC Descriptors" which aimed to adapt established criteria for the evaluation of information retrieval systems to the assessment of searcher performance in online bibliographic searching. The concept of relevance is discussed, and the Cranfield paradigm for the evaluation of information…

These two articles discuss clustering structure in the Cystic Fibrosis Document Collection, which is derived from the National Library of Medicine's MEDLINE file. The exhaustivity of four subject representations and two citation representations is examined, and descriptor-weight thresholds and similarity thresholds are used to compute…

Describes SAPHIRE, an experimental information retrieval system featuring concept-based automated indexing, natural language input, and relevance-based retrieval; and compares its retrieval effectiveness with other indexing approaches in a Boolean-based searching environment. Indexing in MEDLINE is described, and a test collection of AIDS…

A measure of indexing consistency is developed based on the concept of 'fuzzy sets'. It assigns a higher consistency value if indexers agree on the more important terms than if they agree on less important terms. Measures of the quality of an indexer's work and exhaustivity of indexing are also proposed. Experimental data on indexing consistency…

Analyzes relevance assessments of topical descriptors for bibliographic records for two dimensions: (1) a vertical conceptual hierarchy of broad to narrow descriptors, and (2) a horizontal linkage of related terms. The data were analyzed for a semantic distance and semantic direction effect as postulated by the Semantic Distance Model. (Author/LRW)

Proposes a model of knowledge-based information retrieval (KBIR) that is based on a hierarchical concept graph (HCG) which shows relationships between index terms and constitutes a hierarchical thesaurus as a knowledge base. Conceptual distance between a query and an object is discussed and the use of Boolean operators is described. (25…

Summarizes the development of information retrieval evaluation ideas, describes the design of the TREC (Text Retrieval Conference) experiments, and discusses the Okapi team's participation in TREC. Highlights include the Cranfield projects that tested the principles of information retrieval system design, test collections, weighting functions,…

Discussion of improving subject access in online library catalogs focuses on a study that used a book description from "Book Review Digest" to evaluate the information on the bibliographic record found on the OCLC database. Matching by Library of Congress Subject Headings and by keywords in the title are discussed. (22 references) (LRW)

Two articles discuss the clustering of composite representations in the Cystic Fibrosis Document Collection from the National Library of Medicine's MEDLINE file. Clustering is evaluated as a function of the exhaustivity of composite representations based on Medical Subject Headings (MeSH) and citation indexes, and evaluation of retrieval…

An indexer faces unique challenges when the genre consists of biographical material written for children. The narrative prose style demands vividly descriptive indexing terms, but the youthful audience requires a control of vocabulary and a clear, simple presentation. The literature of library and information science yields a few sources on the…

Presents a comparison of collocation-based similarity measures for the proper selection of additional search terms in query expansion. Highlights include evaluating retrieval effectiveness in a vector space model, thesauri, document ranking, and experiments on two Korean test collections. (Author/LRW)

Describes the design and implementation of TACHIR, a tool for the automatic construction of hypertexts for information retrieval. Topics include a three-level conceptual model; navigating among documents, index terms, and concepts; the use of HTML (Hypertext Markup Language) and the World Wide Web; evaluation of TACHIR; and future possibilities.…