This analysis of Zipf's law uses an index for the sequence of observed values of the variables in a Zipf-type relationship. Three important properties relating rank, count, and frequency are identified, shapes of Zipf-type curves are described, and parameters of the Mandelbrot-Zipf law are discussed. (Author/LRW)

An analysis of the question of the optimal depth of indexing in order to design an effective document retrieval system is presented. It is shown that some more fundamental questions about indexing and retrieval rather than indexing depth are central to the issue. (Author/MBR)

Analyzes the nature of Boolean information retrieval in relation to the discrete weights of query terms, examines assigned weights from an approach involving thresholds, and generates an evaluation mechanism which allows the user to attach a threshold to the query term. Thirteen references are listed. (FM)

Introduces a new model of viewing documents based on citing-cited relationship between them. Using Bayes' decision theory, it is shown how source document may be indexed and weighted by relevant cited document features, corresponding to one pass relevance feedback Model 1 (probabilistic indexing) or Model 2 (probabilistic retrieval). (24…

Presents adaptation of a probabilistic theoretical model previously used in relevance feedback for use in automatic indexing of documents (in the sense of imitating) human indexers. Methods for model application are proposed, independence assumptions used in the model are interpreted, and the probability of a dependence model is discussed.…

Highlights semantic links and shows how the semantic content of hypertext links can be used for information retrieval. Discussion includes indexing and retrieval algorithms that exploit link content and node content; retrieval strategies exploiting semantic links, including conventional retrieval and constrained spreading activation techniques;…

Formal proofs are given of the effectiveness under well-defined conditions of the thesaurus method in information retrieval. It is shown, in particular, that when certain semantically related terms are added to the information queries originally submitted by the user population, a superior retrieval system is obtained in the sense that for every…

Describes the two basic approaches to the calculation of term discrimination values for automatic indexing. The results of an experiment that investigated the differences between algorithms of these two approaches in terms of their impact on the discrimination value model are reported and discussed. (13 references) (Author/CLB)

Discusses the effect of including relevance data on the calculation of term discrimination values in bibliographic databases. Algorithms that calculate the ability of index terms to discriminate between relevant and non-relevant documents are described and tested. The results are discussed in terms of the relationship between term frequency and…

Describes three models for probabilistic indexing, all based on the Darmstadt automatic indexing approach, and presents experimental evaluation results for each. The discussion covers the improved retrieval effectiveness of probabilistic indexing over binary indexing, and suggestions for using this automatic indexing method with free text terms.…

Outlines an approach to the automatic construction of a knowledge base resulting in a system that is able in each phase of its construction to acquire domain knowledge from all new information that it is building, particularly index terms. Topics covered include production rules, the use of semantic networks, and user interfaces. (32 references)…

Describes a new method for automatic indexing and retrieval called latent semantic indexing (LSI). Problems with matching query words with document words in term-based information retrieval systems are discussed, semantic structure is examined, singular value decomposition (SVD) is explained, and the mathematics underlying the SVD model is…

The major advantage of sequential classification, a technique for automatically classifying documents into previously selected categories, is that the entire document need not be processed before it is classified. This method assumes the availability of a priori categories, a selection of keywords representative of these categories, and the a…

Summarizes the experimental evidence that indicates that text indexing systems based on the assignment of appropriately weighted single terms produce retrieval results superior to those obtained with more elaborate text representations, and provides baseline single term indexing models with which more elaborate content analysis procedures can be…

The feasibility of using a distance measure, called the Bayesian distance, for automatic sequential document classification was studied. Results indicate that, by observing the variation of this distance measure as keywords are extracted sequentially from a document, the occurrence of noisy keywords may be detected. This property of the distance…