Search results

The problem of automatic diagnosis by use of a computer is expressed as an optimization problem in which parameters are chosen to minimize the diagnosis errors in reference to a previously treated set of patients. The results are expressed in terms of statistical measures of mutual associations of symptoms, and of symptoms with diseases. A decision criterion is discussed, and a formula is derived to describe the diagnostic value of each symptom. No assumptions are made regarding mutual exclusiveness of diseases or statistical independence of symptoms.

This paper provides an introduction to the use of n‐grams in textual information systems, where an n‐gram is a string of n, usually adjacent, characters extracted from a section of continuous text. Applications that can be implemented efficiently and effectively using sets of n‐grams include spelling error detection and correction, query expansion, information retrieval with serial, inverted and signature files, dictionary look‐up, text compression, and language identification.

Federal attempts to stimulate technological innovation have been unsuccessful because of the application of an inappropriate policy framework that lacks conceptual and empirical knowledge of the process of technological innovation and fails to acknowledge the relationship between knowledge production, transfer, and use as equally important components of the process of knowledge diffusion. This article argues that the potential contributions of high‐speed computing and networking systems will be diminished unless empirically derived knowledge about the information‐seeking behavior of the members of the social system is incorporated into a new policy framework. Findings from the NASA/DoD Aerospace Knowledge Diffusion Research Project are presented in support of this assertion.

Thirty years or more ago, a favoured question in examination papers for librarians was some variation on the theme ‘Compare the merits of a classified catalogue and an alphabetical subject catalogue’. This was a subject which it was possible to write on or to discuss at great length, advancing a number of theoretical arguments or opinions expressed by pundits, without ever stating a single demonstrable fact. When, in the early 1950's, various people, such as Taube and Mooers, proposed new techniques for indexing, the reaction from the traditionally‐minded was such that it appeared there was to be a repetition of all the old arguments. In an editorial in American Documentation in 1955, Perry expressed a viewpoint, shared by many others, when he wrote:

Genetic algorithms are a class of non‐deterministic algorithms that derive from Darwinian evolution and that provide good, though not necessarily optimal, solutions to combinatorial problems. We describe their application to the identification of characteristics that occur approximately equifrequently in a database, using two different methods for the creation of the chromosome data structures that lie at the heart of a genetic algorithm. Experiments with files of English and Turkish text suggest that the genetic algorithm developed here can produce results superior to those produced by existing non‐deterministic algorithms; however, the results are inferior to those produced by an existing deterministic algorithm.

My originalthis paper was to take, from the literature, examples of successful solutions of practical problems found in using external information services. This would have been fine if there had been any examples in the literature, but there were none that I found. No one has said in print what happens when tapes are held up for three months in a dock strike, or how much it costs to re‐profile when the thesaurus is revised without consultation with the user, or how much extra work is involved when the record format is changed at short notice. No one has actually stated in public that he allowed two hours per profile and it actually required ten, or that the programming costs were three times as large as his (and his Computer Department Manager's) estimate. These things happen—but one naturally does not admit to them in print, even though by so doing one could be of inestimable value to one's professional colleagues.

The purpose of this study is to examine the growth patterns of tag vocabulary in collaborative tagging systems to verify the sustainability and stabilization of tag distributions. Both sustainability and stabilization are essential to the mining and categorization of information driven by tagging behaviors.

The study was based on time series data of CiteULike from November 2004 to April 2010. Power law distributions were detected to reveal statistical regularities and tagging patterns. Logistic regression analysis with time‐dependent covariates was conducted to identify the factors affecting the growth of distinct tags for articles. The significance of the effects and the time taken for a given article to reach its tagging maturity were also explored.

Time series plots and trend analysis illustrated the continuous growth of the tagging system. Exploratory analysis of power law distribution fittings indicated a sign of system stability known as scale invariance. Logistic regression results demonstrated that for a particular article, the number of users who tagged the article, the initial date when the article was tagged, and the life span of the article are statistically significant to the ratio of the distinct tag number to the total tag number for a given article. These results confirmed that the distinct tag ratio of an article gives rise to a stable pattern.

Though extensive work has been done on the patterns of tag vocabulary, it is not clear how the growth of distinctive tags behaves in relation to the total number of tag applications, considering time‐dependent covariates such as the number of users, and the longevity of an article. This paper sets to complement the literature on the existing methodology and investigate this property in detail.

A study has been made of the effect of controlled variations in indexing vocabulary size on retrieval performance using the Cranfield 200 and 1400 test collections. The vocabularies considered are sets of variable‐length character strings chosen from the fronts of document and query terms so as to occur with approximate equifrequency. Sets containing between 120 and 720 members were tested both using an application of the Cluster Hypothesis and in a series of linear associative retrieval experiments. The effectiveness of the smaller sets is low but the larger ones exhibit retrieval characteristics comparable to those of words.