1 / 20

The three Bibliometric ‘Laws’ – plus one

The three Bibliometric ‘Laws’ – plus one. Peter Ingwersen Royal School of LIS, DK Oslo University College, NO. Three Bibliometric “laws”. Lotka’s law (1926): www.lis.uiuc.edu/~jdownie/biblio/lotka.html

lemonj
Download Presentation

The three Bibliometric ‘Laws’ – plus one

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The three Bibliometric ‘Laws’ – plus one Peter Ingwersen Royal School of LIS, DK Oslo University College, NO

  2. Three Bibliometric “laws” • Lotka’s law (1926): www.lis.uiuc.edu/~jdownie/biblio/lotka.html • Productivity of authors (researchers); how many researchers have written 1, 2, 3… articles? • Frequency distribution (probability) • Bradford’s law (1934): www.lis.uiuc.edu/~jdownie/biblio/bradford.html • Scatter of the articles of a given subject over journals • equal-production groups • Zipf’slov (1949): www.lis.uiuc.edu/~jdownie/biblio/zipf.html • The frequency of words in text • Rank/Frequency distribution 2011

  3. Lotka’s law • Frequency distribution of scientific production • Lotka’s law of authorship describes the publication frequencies for authors within a given domain: • " . . . the number (of authors) making n contributions is about 1/n² of those making one; … and the proportion of all contributors, that make a single contribution, is about 60 percent" (Lotka, 1926) 2011

  4. Lotka’s law • f(n) = C / n2 - where f(n) denotes the number of authors within a domain who, in a given period, have published n articles • C is a constant and denotes the number of researchers that have produced 1 article • The number of researchers that have published n articles is proportional with 1 / n2 – that is … 2011

  5. Lotka’s law • If 100 researchers have published one article each within a given domain then Lotka’s law predicts that: • 100 / 22 = 25 researchers, will have published 2 articles, • 100 / 32 = 11 researchers, will have published 3 articles, and • 100 / 72 = 2 researchers will have published 7 articles … 2011

  6. Lotka further • This van be illustrated (and verified/falsified) by means of Dialog Rank command • In large corpora and over a suitable time period (domain dependent) Lotka’s Law is quite accurate, but not (necessarily) in a statistical manner. 2011

  7. Bradford’s law • Bradford’s law of scattering is often used as a guideline to determine the number of core journals within a given subject • Most of the articles on a given subject have a tendency to be bunched together in a small number of journals • The rest tends to scattered over a large number of journals 2011

  8. Bradford’s law (verbal) • Two different wordings: a verbal and a graphical “If scientific journals are arranged in order of decreasing productivity of articles on a given subject, they may be divided into a nucleus of periodicals more particularly devoted to the subject and several groups or zones containing the same number of articles as the nucleus, when the number of periodicals in the nucleus and succeeding zones will be as 1 : n : n2” (S. C. Bradford, 1934) 2011

  9. Bradford’s law • Bradford’s law is usually presented in equal-production groups form, with this special procedure: • rank the journal-titles in order of their productivity (i.e. in order of the frequency with which they publish articles on the specified subject) • divide the journal-titles into m groups or zones, where each zone produces 1/m of the total number of articles • then the proportion of journals in each zone will be definable by the progression n0:n1:n2 ..., where n is the Bradford multiplier 1 : n : n2 2011

  10. Bradford’s law • e.g. m = 3 zones: • top 8 journals produce 110 articles • next 29 journals produce 133 articles • next 127 journals produce 152 articles • number of journals in each zone follows the progression 8:29:127 ... which is roughly equal to the progression 1:4:17... which is roughly equal to 40:41:42, so n = 4(8x1 : 8x4 : 8x16) 2011

  11. Bradford’s law • The mathematical proportion between the number of journals in the core and in zone 2 is a constant = n (Bradford’s multiplier); and to zone 2 n² • Bradford expressed this relationship as 1:n:n², but he never defined the law mathematically – n was set at approx. 5 • Simple formulation: an 80/20 rule, that is, 20% of the journals in the bibliography covers 80% of the articles in the bibliography on a given subject over a given time period 2011

  12. Bradford’s law • Example: Bradford’s geophysical data: • The number of articles should ideally be the same in the three zones • Bradford’s law is not statistically exact, but is very widespread as a rule-of-thumb a is defined arbitrarily 2011

  13. Bradford’s law 1, n, n2 = 1, 5, 25 Journ. Articles #Jn #Art 258 journals/ 404 articles5*5*9 = 225 journals = 1/3 of articles a 59 journals/ 499 articles5*9 = 45 journals = 1/3 of articles 9 journals/ 429 articles9 journals = 1/3 of articles 2011

  14.  Groos’ drop Bradford’s law (graphical) Zone 3 Linear scale of cumulative number of relevant articles Zone 2 Zone 1 2011 Logarithmic scale of cumulative number of journals

  15. Bradford’s law • Requirements for the bibliography if Bradford’s law is to apply: • Its subject must be well-defined • It must be complete, that is cover all relevant journals and articles • Only a limited number of years should be examined so that all journals have the same chance of contributing with articles • Groos’ drop: may reflect that the bibliography is not complete, or that the data set is flawed – dirty (duplicates)? 2011

  16. Cumulated articles No. of journals (log) Bradford’s ’Law’, graphic version – Dialog search 2011

  17. … plus Garfield’s ’Law of Concentration’, 1971 • Instead of ’bibliometric scatter’ Garfield suggested ’bibliometric concentration’ – rather an axiom: • Each discipline has a small core of journals covering a large portion of articles (Bradford) • The ’long tail’ (”comet tail”) of each discuipline consists of the cores of other disciplines! • Hence, the limited no. of journals in SCI 2011

  18. Zipf’s law • Zipf’s law is used to predict word frequencies in full text • The law states that if you (in a relatively large text corpus): “…list the words occurring within that text in order of decreasing frequency, the rank of a word on that list multiplied by its frequency will equal a constant.” (Zipf, 1949) • r xf = c, where r is the rank number and f the frequency of each term • (happens to describe the linear part of a Bradford curve) • Nor Zipf’s law is statistically exact, but it is extremely valuable for term weighting purposes in search engines with automatic indexing 2011

  19. Zipf’s law 2011

  20. Zipf’s law Zipf’s law (linear and logarithmic) 2011

More Related