In order to identify and explain the reasons for differences in average on-line search rates among terminal installations, the operations and performance of several facilities using the Lockheed DIALOG system for on-line searching of the ERIC data base were studied. Detailed examinations were made of such aspects as the DIALOG system response time…

The following topics are included in this document: (1) evaluation of plans for implementation of new systems; (2) macroevaluation of document retrieval systems; (3) microevaluation of retrospective searching; (4) evaluation of document acquisition functions; evluation of indexing and indexing languages; (5) evaluation of document screening; (6)…

Three specific contributions to the field of information retrieval are presented. The first two describe the establishment of an adaptive, interactive man-machine dialogue that produces a form of unsolicited librarian-like assistance for the user in his selection of index terms to characterize an indexing function. The data set upon which the…

Part One, Automatic Content Analysis contains: "Content Analysis in Information Retrieval;""The 'Generality' Effect and the Retrieval Evaluation for Large Collections;""Automatic Indexing Using Bibliographic Citations" and "Automatic Resolution of Ambiguities from Natural Language Text." Part Two, Automatic…

Four papers are included in Part One of the eighteenth report on Salton's Magical Automatic Retriever of Texts (SMART) project. The first paper: "Content Analysis in Information Retrieval" by S. F. Weiss presents the results of experiments aimed at determining the conditions under which content analysis improves retrieval results as well…

This book deals with properties of vocabularies for indexing and searching document collections; the construction, organization, display, and maintenance of these vocabularies; and the vocabulary as a factor affecting the performance of retrieval systems. Most of the text is concerned with vocabularies for post-coordinate retrieval systems, with…

Describes study that compared terminology indexing and citation indexing to determine the similarity of documents retrieved. Information science articles from 1971 to 1983 in three journals were studied for term co-occurrences. The distinction between relevant and pertinent documents is discussed, and further research topics are suggested. (LRW)

Assesses the impact of current coverage and indexing policies on MEDLINE's retrieval performance in the medical behavioral sciences literature by comparing retrievals in EXCERPTA MEDICA, PSYCHINFO, SCISEARCH, and SOCIAL SCISEARCH. Searches were made with descriptors, cited references, and natural language, and recall, precision, and novelty ratios…

Study of online searching experience in Martin P. Catherwood Library, New York State School of Industrial and Labor Relations at Cornell University, presents results of five years of collecting data on bibliographic files accessed. A ranking of 10 most frequently searched files and commentary on pattern that emerges are provided. (EJS)

This study of 4,506 journal articles reveals that 4.4% of psychology article titles and 33.4% of mathematics article titles contained at least one eponym each (expression consisting of individuals's name plus word denoting idea associated with person). Twenty-two references, sample eponyms and citations, statistical tests, and source journals are…

Describes preliminary investigation of the use of International Computers Limited's Distributed Array Processor (DAP) for parallel searching of large serial files of documents. DAP hardware and software, test collections, measurement of DAP performance, search algorithms, experimental results, and DAP suitability for interactive searching are…

Describes technique for automatic book indexing requiring dictionary of terms with text strings that count as instances of term and text in form suitable for processing by text formatter. Results of experimental application to portion of book text are presented, including measures of precision and recall. Ten references are noted. (EJS)

Empirical comparisons are made of three statistical sampling methods evaluated for estimating total number of relevant documents in collection for given query: simple random sampling, random sampling which assigns unequal selection probabilities to individual documents in collection, and curve fitting and extrapolation. Nine references and…

Briefly reviews research on citation indexing and citation analysis as techniques for determining relationships between documents and presents the results of a citation analysis of 31 library/information science articles in which source of cited work and source of citing work were considered as predictors of relatedness. Included are 24…

Describes the use of FAMULUS, a database management system, to teach library science students at Case Western Reserve University indexing, computerized file construction, and online information retrieval. The special features of FAMULUS and their use in course instruction are outlined and evaluated. A 19-item reference list is attached. (Author/JL)