Reports experiments with a term weighting model incorporating relevance information in which it is assumed that index terms are distributed dependently. It is argued that, if high recall searches are required, relevance feedback based on the modified dependence model may be superior to the Boolean search. (Author/VT)

Describes an investigation of the use of document components, such as terms, sentences, and whole documents, for indexing and retrieval. A number of probabilistic similarity measures based on document components are studied, as well as a new method of handling probability estimations involving small sample sizes. (24 references) (Author/CLB)

A theoretical framework used to derive a weighting mechanism for ranking output documents in Boolean retrieval systems is described in detail. A simple illustrative example is included, and the implications of such an approach are discussed. (64 references) (Author/CLB)

Examines online search results of sample questions in several databases covering linguistics, cultural anthropology, and physical anthropology in order to determine if and where any overlap in results might occur, and which files have greatest number of relevant hits. Search results by database are given for each subject area. (EJS)

This article illustrates similarities between two different types of full-text databases--Superindex, Harvard Business Review/Online--and uses them as arena to demonstrate search and display applications of full-text. The selection of logical operators, full-text search strategies, and keywords and Bibliographic Retrieval Service's Occurrence…

It is shown that automatic search profiles can be derived which perform comparably with corresponding manual profiles. The most efficient automatic profiles, in terms of performance and search time, are those consisting predominantly of single terms enriched with fragments. (Author/PF)

The relative retrieval performances of the CA Integrated Subject File (CAISF), CA Condensates, and a Merged File created from these two data bases were measured. Retrieval performance is reported in terms of recall, precision values, and costs. (Author/JB)

Minimum, maximum, and average lengths and frequency counts for the data elements in the Chemical Substance and General Subject segments of the Standard Distribution Format (SDF) files distributed by Chemical Abstracts Service are presented. (5 references) (Authors)

An evaluation of graph theoretical clusters of index terms which can be extracted from an automatically indexed document collection, and the effects of employing such clusters in automatic document retrieval are described. (19 references) (Author)

Some problems encountered with the ERIC indexing system in efforts to retrieve information from the ERIC files are discussed. (4 references) (Author/NH)

The experimental best match information retrieval system described is based on serial file organization. Documents and queries are characterized by fixed length bit strings (generated by automatic and manual methods) and character-by-character term match is preceeded by bit string search to eliminate documents which cannot satisfy query.…

An iterative method for information retrieval is presented using searchonyms found from previously retrieved set of documents in query expansion. Large operational bibliographic databases are used to simulate behavior of this method which selects most informative features as searchonyms to be used in query reformulation. Ten references are listed.…

Considers patent subject searching capabilities of major online databases, noting patent claims, "deep-indexed" files, test searches, retrieval of related references, multi-database searching, improvements needed in indexing of chemical structures, full text searching, improvements needed in handling numerical data, and augmenting a…

Introduces methods designed to reduce role of search intermediaries by generating Boolean search formulations automatically using term frequency considerations from natural language statements provided by system patrons. Experimental results are supplied and methods are described for applying automatic query formulation process in practice.…

Addresses problem of developing machine-based information system capable of indicating one document, out of a set of documents written in natural language, which shows maximal degree of homeosemy (relationship in meaning between two texts language-fragments ) toward natural language query. Linguistic aspects, information traces, trigrams,…