STIMULATE 5

Ronald Rousseau Web page: users.telenet.be/ronald.rousseau

Citation analysis

• Advanced parts Bibliographic coupling Co-citation Citation context analysis • The Journal Citation Reports (JCR) Bibliometric indicators

Bibliographic coupling and co-citation

Co-citation network

Co-citation measures

If A is the set of papers citing document X, and B is the set of papers citing document Y, then

A



B

is the set of documents that cite both X and Y, i.e. that co-cite X and Y.

The number of elements in

A



B

, denoted as #

A



B

, is the co-citation frequency of X and Y.

#(

A

#(

A B

) is the relative co-citation frequency

B

)

Citation context analysis

• Cited documents become symbols for the ideas they contain.

• Highly cited documents can be considered as exemplars or concept symbols: illustrations of methods or theories which comprise the essential repertoire of techniques in a specialty.

Citation analysis of scientific journals

• The Journal Citation Reports (JCR) • The annual publication of the JCR by ISI (nowadays Thomson-ISI) has led to a whole series of indicators. Of these the journal impact factor is the best known.

The publication – citation table

Publ year # Publ.

# cit rec’d in 2000 # cit rec’d in 2001 # cit rec’d in 2002 # cit rec’d in 2003 # cit rec’d in 2004 2000 35 5 10 15 15 10 2001 35 5 10 20 12 2002 40 5 12 16 2003 40 6 10 2004 45 8

The ISI impact factor (synchronous) Publ year # Publ.

# cit rec’d in 2000 # cit rec’d in 2001 # cit rec’d in 2002 # cit rec’d in 2003 # cit rec’d in 2004 2000 35 5 10 15 15 10 2001 2002 35 5 10 20 12 40 5 12 16 2003 2004 40 6 10 45 IF(2003) = (20+12)/ (35+40) =0.427

8

Mathematical formulas: the general synchronous impact factor (Rousseau, 1988) CIT(Y-2,Y) + CIT(Y-1,Y) IF(Y)= --------------------------------- = IF 2 (Y) PUB(Y-2) + PUB(Y-1)

n s



s n

1 

k s s n

1   

k

)

Diachronous impact factors

Publ year # Publ.

# cit rec’d in 2000 # cit rec’d in 2001 # cit rec’d in 2002 # cit rec’d in 2003 # cit rec’d in 2004 2000 35 5 10 15 15 10 2001 35 5 10 20 12 2002 40 5 12 16 2003 40 6 10 2004 45 DIF 3 (2002) = (5+12+16) / 40 = 0.825

8

Mathematical formulas: the general diachronous impact factor

n s



s n

1  

k

)

The source – item table

Period of origin Period 1 Period 2 Period 3 Period 4 Period 5 # sources # items period 1 # items period 2 # items period 3 # items period 4 # items period 5 30 5 10 15 15 10 35 5 10 20 12 40 5 12 16 40 6 10 45 8

Examples of general source-item relations • Scientific journal  • Scientific journal  (publishes) articles (publishes) issues   (receives) citations (receives) citations • Scientific journal  (publishes) one particular issue citations • Congress proceedings  (consists of) articles   (receives) (receives) citations • Institute  (publishes) webpages  (receives) inlinks (the institute web impact factor) • Country  (publishes) webpages countries) (country web impact factor) • Fiction author   (receives) inlinks (from other (writes/publishes) books  (realizes) sales • Country  (publishes) articles in a particular domain  citations • Journal  (publishes) articles  particular journal (receives) (receives) citations in one

Publ year # Publ.

# cit rec’d in 2000 # cit rec’d in 2001 # cit rec’d in 2002 # cit rec’d in 2003 # cit rec’d in 2004

Generalized impact factors (Frandsen – Rousseau)

2000 2001 2002 2003 2004 35 5 10 15 15 10 35 5 10 20 12 40 5 12 16 40 6 10 45 8

The median impact factor (MIF): a new impact indicator



J J J

 Here TOT denotes the total number of citations received by journal J in the year Y. CPUB(Y-X,Y) denotes the cumulative number of publications in the journal J, during the period [Y-X, Y]. The symbol X denotes the median cited age.

References

SOMBATSOMPOP, N., MARKPIN, T., PREMKAMOLNETR, N. (2004), A modified method for calculating the impact factors of journals in ISI Journal Citation Reports: polymer science category in 1997-2001.

Scientometrics

, 60: 217-235.

ROUSSEAU, R. (2005), Median and percentile impact factors: a set of new indicators.

Scientometrics

, 63: 431-441.

Example: Scientometrics MIF(2003) • TOT(2003) = 1012  TOT/2 = 506 • The median cited age is 6.53 • The number of articles published during these 6.53 years is 591.75

• The 2003 MIF of

Scientometrics

is 506/591.75 = 0.855 < ISI IF (1.251).

Y 03 02 01 00 99 98 97 Publ – cum.sum

83 Cit 22 84-167 96 91-258 82 - 340 128-468 84-552 75-627 Cum cit % 2.17

11.66

123 23.81

85 32.21

68 78 64 38.93

46.64

52.96

Recent trends

Not just the ISI 2-year synchronous impact, but a whole battery of impact factors are used in science evaluation.

What should be the real purpose of research evaluation?

The real goal of any form of research evaluation is providing those people and institutions that have the talent and motivations to carry out scientific research, with the best conditions possible under which to do so (Russell-Rousseau, 2002).

Evaluation and scientometric research

Budgetary and other kinds of constraints make evaluations necessary for the equitable distribution of resources. The evaluation of short-term strategic research as well as long-term curiosity-driven search for new knowledge demands the same rigorous standards as scientific research itself. For this reason the application of bibliometric and scientometric techniques in research evaluation must keep up with the rapid changes occurring in scientific communication patterns.

Information scientists must also constantly improve the theoretical foundation for the construction of output and impact indicators supporting peer review (Russell-Rousseau, 2002)