www.uhasselt.be
DSpace

Document Server@UHasselt >
Research >
Research publications >

Please use this identifier to cite or link to this item: http://hdl.handle.net/1942/3532

Title: Generalized success-breeds-success principle leading to time-dependent informetric distributions
Authors: EGGHE, Leo
ROUSSEAU, Ronald
Issue Date: 1995
Publisher: JOHN WILEY & SONS INC
Citation: JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE, 46(6). p. 426-445
Abstract: The success-breeds-success principle (SBS principle) is reformulated in order to generate a general theory of source-item relationships. Several extensions are included such as a time-dependent probability alpha(t) for a new source to enter the system and general probabilities for new items to be produced by an old source. Moreover, we allow non steady-state situations. A new model for the expected probability E(P(t, n))-the expectation of the fraction of sources having n items at time t-is presented and compared with the SBS principle. As these models involve mathematically prohibited approximations, they are both compared with exact combinatorial calculations. Criteria for E(P(t, n)) to be decreasing (or not) in t as well as in n are given. It is observed that, even in the classical SBS framework, distributions which are not strictly decreasing in n are commonly encountered. Finally, introducing a quasi steady-state assumption, we show that nearly all classical frequency distributions, such as the beta function, the Lotka distribution, the truncated geometric distribution and the truncated Poisson distribution can be derived and explained via this generalized SBS principle. Note, however, that our models lead to time-dependent versions of these classical distributions.
Notes: UNIV INSTELLING ANTWERP,INFORMATIE BIBLIOTHEEKWETENSCHAP,B-2610 WILRIJK,BELGIUM. KIHWV,B-8400 OOSTENDE,BELGIUM.EGGHE, L, LIMBURGS UNIV CENTRUM,UNIV CAMPUS,B-3590 DIEPENBEEK,BELGIUM.
URI: http://hdl.handle.net/1942/3532
DOI: 10.1002/(SICI)1097-4571(199507)46:6<426::AID-ASI3>3.0.CO;2-B
ISI #: A1995RG04400002
ISSN: 1532-2882
Type: Journal Contribution
Appears in Collections: Research publications

Files in This Item:

Description SizeFormat
Published version1.52 MBAdobe PDF

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.