Article
Weighted Impact Factor (WIF) for assessing the quality
of scientific journals
Rasim Alguliyev
Institute of Information Technology, Azerbaijan National Academy of Sciences, Azerbaijan
Ramiz Aliguliyev
Institute of Information Technology, Azerbaijan National Academy of Sciences, Azerbaijan
Nigar Ismayilova
Institute of Information Technology, Azerbaijan National Academy of Sciences, Azerbaijan
Abstract
Nowadays impact factor is the significant indicator for journal evaluation. In impact factor’s calculation is used number of all
citations to journal, regardless of the prestige of cited journals, however, scientific units (paper, researcher, journal or scientific
organization) cited by journals with high impact factor or researchers with high Hirsch index are more important than objects cited
by journals without impact factor or unknown researcher. In this paper was offered weighted impact factor for getting more
accurate rankings for journals, which consider not only quantity of citations, but also quality of citing journals. Correlation
coefficients among different indicators for journal evaluation: impact factors by Thomson Scientific, weighted impact factors offered
by different researchers, average and medians of all citing journals’ impact factors and 5-year impact factors were analysed.
Keywords
Impact factor; weighted impact factor; citing journals; citation counting
1. Introduction
Today, a lot of scientific papers in different subject areas, fields are published. So determining the influence of each
paper or journal where these papers are published is very important. Nowadays the most used indicator for journal
influence measuring is impact factor proposed by Garfield [1]. The generally accepted as an indicator of journal prestige
the Thomson Reuters (earlier Institute of Scientific Information - ISI) calculates the impact factor of jth journal in year t
as ratio of number of citations in current year to articles published in this journal during previous two years to number of
these articles:
∑
(1)
where njt – is the number of journals in the set of all journals indexed in year t cited the articles of journal, cijt – is the
number of citations to journal j from journal i in year t and ajt – is the total number of articles published in year t.
Although IF is an important quotient for evaluation of scientific journals used by librarians, researchers, science
policymakers, there are many critiques against the IF efficiency [2-3]. Reedjik coordinates it by wrong, biased and even
manipulated citations, as a result of citation habits for authors in different fields, selectivity in citations by authors,
errors made by authors in citation lists at the end of documents and by ISI in entering publications and citations in
databases, in classifying citations and accrediting them to journals and authors [4]. During the last decades were offered
Corresponding author:
Ramiz Aliguliyev, Institute of Information Technology, Azerbaijan National Academy of Sciences, 9 B. Vahabzade Street, AZ1141, Baku, Azerbaijan
E-mail: r.aliguliyev@gmail.com
Alguliyev et al.
2
different modifications of ISI IF. For using with complex citation databases as Scopus was proposed the SCImago
Journal Rank on citation weighting schemes and eigenvector centrality was proposed [5]. Perez-Hornero et al. proposed
a Bayesian approach to the problem with taking into account journals’ recent trajectory besides the current prestige of
given journal [6]. Although during the last decades were proposed new indicators for journals evaluation, was
concluded that, they (the Eigenvector, the SCImago journal rank, an Article Influence Score and journal h-index)
correlate very much with the IF proposed by ISI and among each other [7]. One of the main disadvantages of IF is the
equality of citations, regardless of the importance of the citing journals [8-10]. Unfortunately, in calculation of IF
prestige, reputation of citing journals are not considered, nevertheless 10 citations from journal with high IF must be
preferred to by one citation from 10 journals with low IF or without IF. For solution of this problem in this paper was
proposed impact factor calculated by taking into 5-year impact factor of each citing journal.
2. Related Work
In general, journals in specialized and applied disciplines get more citations than journals in fundamental subject areas.
The numbers of researchers and publishing journals in each area greatly affect the impact factor [11]. In this reason
some researchers proposed methods for normalization of impact factors of journals for different scientific areas. Owlia
et al. introduced normalized impact factor for evaluating the quality of journals and research works in different
disciplines [12]. The normalized IF (NIF) index was established based on multiplication of journal IF by constant factor.
NIF is the ratio of the aggregate impact factors of journals to aggregate impact factor of each discipline in given
scientific field:
̂
(2)
where Ct - is the total number of citations to the entire given scientific field in current year, At - is the total number of
articles published by journals from entire given scientific field in current year, Ĉt- is the total number of citations to the
journals of the particular discipline and Jt - is the total number of journals of the particular discipline published articles.
Other method for normalization of journal impact factors was produced by Iftikhar, Masood and Song [13]. Modified
impact factors were calculated at disciplines, branches and specialities level and designated as Red, Yellow and Green
MIF respectively. For modification the highest ISI IF of journal of that discipline, branch or speciality is weighted as
100 and other groups members are normalized accordingly by considering it as reference point. Also was proposed
rank-normalized impact factor [14], Source Normalized Impact per Paper (SNIP) by Scopus [15].
Weighting citations for more effective impact factor and more accurate ranking of journals have also been suggested by
some researchers. In 1976 was developed self-consistent methodology for determining citation influence weights for
scientific journals, fields and subfields. Also was proposed modification of IF using Weighted PageRank [16]. Authors
suggested a simple combination of both the IF and the weighted PageRank and found that the resulting journal rankings
correspond well to a general understanding of journal status [17]. Zyezkowski formulated weighted impact factor of
journals using weighted citations, weighted impact factor of papers, weighted Hirsch index and weighted efficiency
index [18]. In order to not ignoring the impact or prestige of the cited journals were proposed different modifications of
impact factor. Buela-Casal had offered two indices: a mean impact factor of the citing journals (MIFCJ) and a weighted
impact factor (WIF). MIFCJ is quotient of multiplication of impact factors of citing journals by the number of cited
articles to the total number citing articles [19]:
∑
(3)
Weighted impact factor of a journal is the average of MIFCJ and IF from ISI Journal Citation Ranking (JCR) reports:
(4)
Habibzadeh and Yadollahie suggested weighted impact factor with relative weights of journals by their IF in the
previous year normalized using logistic function [20]:
Alguliyev et al.
3
∑
where
(5)
designates the weight of relative to Ji weighted impact factor of journal Jk in year t:
(6)
where
- is a quotient the numerator of which is the IF of citing journal in the previous year and the denominator is IF
of cited journal in the previous year:
(7)
Although proposed by researchers weighted IF takes into account not only the number of times that the journal has been
cited, but also the prestige of the journals by which it has been cited, were determined some serious problems of
proposed WIF (weighted impact factor). Because of ranking of journals based in this WIF can be misleading, was
indicated how the problems with the WIF proposed by Habibzadeh and Yadollahie (H&Y) can be solved [21].
In our opinion, the main lack of WIF proposed by H&Y is dependence on IF of cited journal in previous year.
Example: let us calculate weighted impact factors of 3 journals for current year with the same number of articles in
previous two years, we assume that number of citations and impact factors of citing journal to these articles in current
year are the same for each journal (table 1):
Number of articles: Pi = 20;
Impact factors of journals in previous year: IF1 = 2; IF2 = 4; IF3 = 6;
Table 1. Impact factors and numbers of citations of citing journals for journal in example.
N
IF of citing journal
Number of citations
1
2
3
4
2
4
8
16
2
2
2
2
Using data from table 1 were calculated weighted impact factors for given journals proposed by H&Y:
WIF1 = 0.994324; WIF2 = 0.759668; WIF3 = 0.443629
As result we can see that, calculated weighted impact factors for 3 journals with the same number of citations from the
same sources are different depending on impact factors of journals in previous year. Cited journals with high impact
factor in previous year gets lower weighted impact factors. This disadvantage of proposed IF makes ranking of journals
by this method undesirable.
3. Proposed weighted impact factor
Assume that St – is the set of all journals indexed in year t, nt – is the number of indexed journals, njt – is the number of
journals in St cited the articles of journal j, ajt – is the total number of articles published in year t, cijt – is the number of
citations to journal j from journal i in year t, FIFjt – is 5-year impact factor of journal j in year t. Proposed weighted IF
of journal j in year t is
∑
(
)
(7)
To analyse the efficiency of proposed impact factor, we tried to calculate new impact factors in 2013 for selected twenty
journals in computer science sphere from JCR for 2013 (Table 2).
Table 2. Some indicators of analysed journals from Scopus Database.
Alguliyev et al.
4
N
Journal
Number of articles
published in 2011 and
2012
Number of citations in
2013 to articles
published in 2011 and
2012
Impact factor in
2012
1
2
3
4
5
6
7
8
Neural Computation
Swarm Intelligence
Neural Processing Letters
Artificial Life
Cognitive Computation
Computer Speech And Language
Fuzzy Optimization and Decision Making
Genetic Programming and Evolvable
Machines
International Journal of Appled Mathematics
and Computer Science
Journal of Ambient Intelligence and Smart
Environments
ACM Transactions on Applied Perception
ACM Transactions on Knowledge Discovery
from Data
Acm Transactions on Information Systems
ACM Transactions on the Web
ACM Transactions on Sensor Networks
Acm Transactions on Software Engineering
And Methodology
IEEE Transactions on Computational
Intelligence and AI in Games
IEEE Transactions on Dependable and
Secure Computing
IEEE Transactions on Autonomous Mental
Development
World Wide Web
226
26
76
48
88
67
45
42
383
48
94
93
97
121
45
45
1.76
0.64
1.24
1.585
0.867
1.463
1.488
1.333
136
189
1.008
74
80
1.298
40
37
42
42
1
1.676
42
39
54
37
55
62
79
54
1.07
1.405
1.444
1.548
50
58
1.694
143
163
1.059
54
73
2.17
58
94
1.196
9
10
11
12
13
14
15
16
17
18
19
20
For these journals number of articles published in 2011 and 2012, total number of citations in 2013, number of
sources citing journal were determined using data about articles and their citations from Scopus database. Also the
numbers of citations from each journal with impact factors were determined for given journals (Table 3).
Table 3. 5-year impact factors of citing journals and number of citations from them.
J1 – Neural Computation
1
35.89 (1)
10
13.45 (1)
19
6.144 (5)
28
4.25 (1)
37
3.668 (2)
46
3.108 (1)
55
2.733 (1)
64
2.38 (31)
73
1.938 (3)
82
1.6 (1)
2
34.37 (1)
11
11.34 (3)
20
6 (1)
29
4.24 (11)
38
3.646 (1)
47
3.069 (1)
56
2.653 (2)
65
2.339 (1)
74
1.922 (3)
83
1.596 (1)
3
31.03 (3)
12
10.58 (9)
21
5.94 (14)
30
4.049 (7)
39
3.632 (9)
48
3.068 (3)
57
2.61 (20)
66
2.307 (3)
75
1.871 (1)
84
1.595 (1)
4
23.17 (1)
13
10.45 (1)
22
5.484 (1)
31
4.017 (1)
40
3.612 (2)
49
3.05 (1)
58
2.567 (1)
67
2.287 (3)
76
1.811 (8)
85
1.572 (1)
5
16.41 (1)
14
9.924 (2)
23
4.885 (1)
32
3.879 (2)
41
3.607 (8)
50
2.998 (3)
59
2.526 (1)
68
2.27 (1)
77
1.767 (1)
86
1.55 (1)
6
16.40 (4)
15
7.87 (10)
24
4.544 (1)
33
3.844 (1)
42
3.568 (1)
51
2.927 (2)
60
2.525 (1)
69
2.158 (1)
78
1.745 (1)
87
1.529 (2)
7
13.58 (1)
16
7.463 (2)
25
4.479 (4)
34
3.71(1)
43
3.291 (1)
52
2.895 (1)
61
2.5 (10)
70
2.143 (2)
79
1.732 (1)
88
1.402 (2)
8
14.57 (1)
17
7.063 (4)
26
4.422 (1)
35
3.707 (1)
44
3.219 (1)
53
2.892 (4)
62
2.496 (1)
71
2 (1)
80
1.724 (1)
89
1.386 (2)
9
14.47 (1)
18
6.895 (3)
27
4.284 (5)
36
3.676 (1)
45
3.146 (1)
54
2.743 (1)
63
2.484 (2)
72
1.947 (1)
81
1.643 (1)
90
1.338 (1)
Alguliyev et al.
91
1.336 (1)
100
1.14 (3)
109
0.548 (1)
5
92
1.319 (1)
101
1.074 (1)
110
0.297 (1)
J2 – Swarm Intelligence
1
6.226 (1)
10
1.364 (1)
2
5.165 (2)
11
0.953 (2)
93
1.314 (1)
102
1.032 (1)
111
0.187 (2)
94
1.305 (1)
103
0.97 (1)
95
1.231(1)
104
0.84 (1)
96
1.216 (1)
105
0.816 (1)
97
1.192 (1)
106
0.697 (1)
98
1.183 (1)
107
0.672 (2)
99
1.182 (1)
108
0.6 (1)
3
3.448 (1)
4
3.405 (1)
5
3.097 (2)
6
2.747 (1)
7
1.957 (1)
8
1.615 (1)
9
1.545 (2)
3
3.676 (1)
12
1.831 (1)
21
1.216 (1)
30
0.753 (1)
4
3.632 (2)
13
1.811 (8)
22
1.183 (2)
31
0.716 (1)
5
3.513 (1)
14
1.71 (1)
23
1.074 (6)
32
0.682 (5)
6
3.219 (1)
15
1.529 (1)
24
1.04 (3)
33
0.64 (1)
7
2.501 (2)
16
1.454 (2)
25
0.898 (1)
34
0.622 (1)
8
2.457 (3)
17
1.42 (1)
26
0.866 (1)
35
0.561 (1)
9
2.384 (2)
18
1.36 (1)
27
0.84 (2)
36
0.497 (2)
3
7.51 (1)
12
2.496 (1)
21
1.336 (1)
4
7.435 (1)
13
2.333 (1)
22
0.953 (2)
5
6.69 (1)
14
2.307 (1)
23
0.816 (1)
6
6.226 (1)
15
2 (1)
24
0.617 (1)
7
5.165 (2)
16
1.945 (1)
25
0.48 (1)
8
4.728 (1)
17
1.777 (1)
26
0.417 (1)
9
4.446 (1)
18
1.545 (2)
4
4.479 (1)
13
2.538 (1)
22
1.811 (2)
31
0.735 (1)
5
4.372 (1)
14
2.525 (1)
23
1.745 (1)
32
0.592 (1)
6
4.244 (3)
15
2.501 (3)
24
1.596 (1)
33
0.326 (1)
7
4.017 (1)
16
2.445 (1)
25
1.529 (2)
8
3.674 (1)
17
2.339 (1)
26
1.52 (1)
9
3.598 (1)
18
2.194 (1)
27
1.423 (1)
4
2.339 (2)
13
1.137 (4)
22
0.466 (1)
5
1.952 (2)
14
1.074 (2)
23
0.305 (1)
6
1.936 (1)
15
0.977 (1)
7
1.915 (2)
16
0.959 (1)
8
1.708 (1)
17
0.932 (1)
9
1.52 (8)
18
0.767 (1)
5
1.81 (15)
14
0.612 (1)
6
1.721 (1)
15
0.27 (16)
7
1.674 (2)
8
1.579 (2)
9
1.386 (1)
6
1.811 (1)
7
1.795 (1)
8
1.726 (2)
9
1.625 (1)
7
2.457 (1)
16
1.529 (2)
25
1.183 (2)
8
2.382 (1)
17
1.504 (1)
26
1.182 (1)
9
2.255 (1)
18
1.454 (1)
27
1.158 (1)
J3 – Neural Processing Letters
1
7.854 (1)
10
2.339 (1)
19
1.329 (1)
28
0.774 (5)
2
4.268 (1)
11
2.143 (1)
20
1.23 (10)
29
0.755 (1)
J4 – Artificial Life
1
17.72 (1)
10
4.406 (1)
19
1.454 (1)
2
13.56 (1)
11
4.244 (7)
20
1.364 (1)
J5 –Cognitive Computation
1
9.924 (1)
10
3.262 (1)
19
2.158 (1)
28
1.157 (2)
2
7.869 (1)
11
2.998 (2)
20
1.938 (2)
29
1.14 (14)
3
7.298 (1)
12
2.847 (1)
21
1.936 (1)
30
0.846 (1)
J6 – Computer Speech And Language
1
7.694 (2)
10
1.423 (5)
19
0.664 (1)
2
2.643 (1)
11
1.41 (1)
20
0.617 (1)
3
2.395 (2)
12
1.146 (1)
21
0.505 (1)
J7 – Fuzzy Optimization and Decision Making
1
3.676 (3)
10
1.364 (4)
2
2.218 (2)
11
1.183 (2)
3
2.167 (1)
12
0.846 (2)
4
2.165 (1)
13
0.746 (1)
J8 – Genetic Programming and Evolvable Machines
1
3.676 (1)
10
1.39 (1)
2
3.027 (1)
11
1.349 (1)
3
2.526 (1)
12
1.282 (8)
4
2.501 (1)
13
1.231 (1)
5
1.938 (1)
J9 – International Journal of Appled Mathematics and Computer Science
1
4.244 (1)
10
2.151 (1)
19
1.368 (2)
2
3.676 (1)
11
2.04 (4)
20
1.364 (2)
3
3.601(1)
12
1.758 (1)
21
1.359 (1)
4
3.212 (1)
13
1.674 (1)
22
1.289 (2)
5
2.64 (2)
14
1.651 (1)
23
1.216 (3)
6
2.62 (1)
15
1.625 (1)
24
1.201 (2)
Alguliyev et al.
28
1.15 (52)
37
0.548 (1)
6
29
1.024 (1)
38
0.483 (1)
30
0.932 (1)
39
0.436 (1)
31
0.898 (2)
40
0.41 (1)
32
0.829 (1)
41
0.395 (1)
33
0.8 (4)
42
0.37 (1)
34
0.671 (1)
43
0.269 (1)
35
0.61 (1)
36
0.594 (3)
5
2.339 (1)
6
2.003 (1)
7
1.947 (1)
8
1.811 (2)
9
1.64 (16)
5
2.61 (1)
14
1.269 (5)
6
2.566 (1)
15
1.216 (1)
7
2.445 (2)
16
1.112 (1)
8
2.395 (1)
17
0.675 (1)
9
2.292 (1)
18
0.5 (1)
5
3.676 (3)
14
0.739 (1)
6
3.371 (1)
15
0.707 (1)
7
3.263 (1)
8
3.068 (1)
9
2.426 (1)
4
2.566 (1)
5
2.446 (1)
6
2.339 (2)
7
1.745 (1)
8
1.716 (3)
9
1.586 (1)
4
3.371 (1)
13
1.452 (1)
5
3.037 (2)
14
1.388 (2)
6
2.927 (1)
15
1.384 (1)
7
2.446 (5)
16
1.322 (2)
8
2.424 (2)
17
1.109 (1)
9
2.339 (3)
18
0.943 (2)
5
2.747 (1)
14
1.169 (1)
6
2.485 (1)
15
1.092 (1)
7
2.395 (1)
16
1.002 (1)
8
2.203 (1)
17
0.765 (1)
9
1.957 (1)
18
0.605 (1)
7
1.322 (1)
16
0.336 (2)
8
1.167 (4)
17
0.269 (1)
9
1.867 (1)
18
0.268 (1)
6
1.282 (1)
7
0.954 (1)
8
0.832 (1)
9
0.723 (3)
6
2.744 (1)
15
1.42 (1)
24
1.093 (3)
33
0.701 (1)
42
0.297 (1)
7
2.426 (1)
16
1.39 (1)
25
1.019 (1)
34
0.697 (3)
43
0.187 (1)
8
2.339 (2)
17
1.341 (1)
26
0.954 (2)
35
0.625 (2)
9
2.259 (4)
18
1.322 (2)
27
0.945 (1)
36
0.606 (1)
J10 – Journal of Ambient Intelligence and Smart Environments
1
3.382 (1)
10
1.529 (1)
2
2.7 (1)
11
1.169 (1)
3
2.632 (1)
4
2.525 (1)
J11 – ACM Transactions on Applied Perception
1
6.144 (1)
10
2.007 (1)
19
0.453 (1)
2
4.283 (2)
11
2 (1)
3
4.017 (1)
12
1.905 (1)
4
2.62 (1)
13
1.36 (2)
J12 – ACM Transactions on Knowledge Discovery from Data
1
4.395 (1)
10
2.339 (3)
2
4.244 (1)
11
1.838 (1)
3
4.017 (1)
12
1.811 (1)
4
3.959 (1)
13
1.359 (2)
J13 – ACM Transactions on Information Systems
1
3.676 (1)
10
1.318 (1)
2
3.371 (3)
11
1.109 (2)
3
3.037 (2)
12
0.466 (1)
J14 – ACM Transactions on the Web
1
7.854 (2)
10
2.158 (3)
19
0.785 (1)
2
4.395 (1)
11
2.033 (1)
20
0.765 (1)
3
3.676 (2)
12
1.469 (1)
21
0.605 (1)
J15 – ACM Transactions on Sensor Networks
1
6.348 (2)
10
1.859 (1)
19
0.43 (5)
2
6.146 (1)
11
1.758 (2)
3
3.587 (2)
12
1.227 (1)
4
3.371 (1)
13
1.183 (4)
J16 – ACM Transactions on Software Engineering And Methodology
1
3.612 (1)
10
0.819 (1)
2
3.371 (1)
11
0.785 (1)
3
2.063 (3)
12
0.727 (1)
4
2.031 (2)
13
0.721 (1)
5
1.756 (1)
14
0.682 (2)
6
1.692 (2)
15
0.43 (1)
J17 – IEEE Transactions on Computational Intelligence and AI in Games
1
3.212 (1)
2
3.071 (1)
3
2.339 (2)
4
1.936 (1)
5
1.63 (11)
J18 – IEEE Transactions on Dependable and Secure Computing
1
6.895 (1)
10
2.067 (1)
19
1.317 (1)
28
0.874 (1)
37
0.605 (1)
2
3.676 (4)
11
2.021 (1)
20
1.291 (1)
29
0.867 (2)
38
0.497 (2)
3
3.371 (2)
12
1.894 (1)
21
1.234 (1)
30
0.819 (1)
39
0.463 (2)
4
3.191 (1)
13
1.726 (10)
22
1.227 (1)
31
0.793 (1)
40
0.43 (3)
5
3.071 (3)
14
1.576 (2)
23
1.106 (1)
32
0.712 (1)
41
0.395 (2)
Alguliyev et al.
7
J19 – IEEE Transactions on Autonomous Mental Development
1
4.244 (4)
10
1.529 (1)
2
3.587 (1)
11
1.423 (1)
3
2.501 (4)
12
1.137 (1)
4
2.26 (17)
13
1.043 (1)
5
2.202 (1)
14
0.898 (1)
6
1.947 (1)
15
0.817 (1)
7
1.859 (1)
16
0.675 (1)
8
1.615 (3)
17
0.567 (1)
9
1.545(2)
1
2
3
3.676 (1)
3.219 (2)
3.191 (1)
10
11
12
1.838 (1)
1.586 (1)
1.45 (23)
here (*) denotes the number of citations
4
2.446 (1)
13
1.251 (1)
5
2.426 (1)
14
1.169 (1)
6
2.403 (1)
15
0.94 (1)
7
2.339 (2)
16
0.332 (1)
8
2.031 (1)
17
0.302 (1)
9
1.955 (1)
18
0.174 (1)
J20 – World Wide Web
Weighted impact factors of given journals were compared with JCR IF for 2013, WIF proposed by H&Y, BuelaCasal and impact factors weighted by average and median of citing journals’ impact factors and 5-year impact factors
(Table 4).
Table 4. JCR IF, WIF by H&Y, by Buela-Casal, proposed WIF, average and median WIF, weighted by average and median of 5-year
impact factors for analysed journals.
N
Journal
JCR IF
WIF by
H&Y
WIF by
BuelaCasal
Propose
d WIF
1
Neural
Computation
1.694 [4]
4.200 [2]
3.603 [1]
7.544 [1]
6.533 [1]
4.200[2]
7.725 [1]
4.9 [3]
2
Swarm
Intelligence
1.833 [2]
4.457 [1]
1.577 [9]
3.472 [4]
3.238 [3]
2.055 [8]
5.204 [3]
5.071 [2]
3
Neural
Processing
Letters
1.237 [12]
1.631 [12]
1.455 [13]
2.763 [8]
1.765 [14]
1.534 [15]
2.02 [13]
1.523 [17]
4
5
Artificial Life
Cognitive
Computation
Computer
Speech And
Language
Fuzzy
Optimization and
Decision Making
Genetic
Programming
and Evolvable
Machines
International
Journal of
Applied
Mathematics and
Computer
Science
Journal of
Ambient
Intelligence and
Smart
Environments
1.93 [1]
1.1 [16]
3.028 [3]
2.649 [4]
2.688 [2]
1.317 [17]
4.762 [2]
2.594 [11]
6.474 [2]
2.253 [9]
4.518 [1]
1.955 [10]
7.724 [2]
2.631 [11]
8.223 [1]
1.996 [12]
1.812 [3]
1.915 [8]
1.778 [7]
2.875 [6]
2.137 [11]
2.317 [4]
3.091 [6]
2.57 [8]
1 [20]
0.855 [20]
1.330 [16]
2.158 [15]
1.172 [20]
1.124 [20]
1.33 [20]
1.48 [20]
1.065 [18]
1.193 [18]
1.363 [15]
1.919 [18]
1.729 [15]
1.453 [16]
1.88 [17]
1.508 [18]
1.39 [9]
1.698 [11]
1.266 [19]
2.443 [12]
1.643 [18]
1.401 [18]
1.815 [18]
1.593 [16]
1.082 [17]
1.198 [17]
1.294 [18]
1.758 [20]
1.658 [17]
1.403 [17]
2.006 [14]
1.773 [14]
ACM
Transactions on
Applied
Perception
1.051 [19]
2.021 [7]
1.264 [20]
2.426 [13]
1.821 [12]
1.700 [12]
2.223 [12]
2.05 [11]
6
7
8
9
10
11
Average
WIF
Median
WIF
Average
5WIF
Median
5WIF
Alguliyev et al.
8
ACM
1.147 [14]
Transactions on
Knowledge
Discovery from
Data
ACM
1.3 [11]
13
Transactions On
Information
Systems
ACM
1.595 [6]
14
Transactions on
the Web
ACM
1.463 [8]
15
Transactions on
Sensor
Networks
ACM
1.472 [7]
16
Transactions On
Software
Engineering And
Methodology
IEEE
1.167 [13]
17
Transactions on
Computational
Intelligence and
AI in Games
IEEE
1.137 [15]
18
Transactions on
Dependable and
Secure
Computing
IEEE
1.348 [10]
19
Transactions on
Autonomous
Mental
Development
World Wide
1.623 [5]
20
Web
[*] denotes the rank of journals.
12
1.573 [13]
1.806 [6]
2.61 [9]
2.480 [6]
2.105 [7]
3.099 [5]
3.118 [5]
2.059 [6]
1.389 [14]
2.3 [14]
2.406 [7]
2.132 [6]
2.897 [9]
3.063 [6]
2.076 [5]
1.829 [5]
3.828 [3]
2.670 [4]
2.234 [5]
3.855 [4]
3.718 [4]
1.697 [10]
1.615 [8]
2.607 [10]
2.285 [8]
1.612 [14]
3.013 [7]
2.184 [10]
1.369 [16]
1.487 [12]
2.098 [16]
1.674 [16]
1.660 [13]
1.912 [15]
1.707 [15]
1.139 [19]
1.549 [10]
1.88 [19]
1.804 [13]
1.965 [9]
1.899 [16]
1.885 [13]
1.533 [14]
1.195 [4]
1.985 [17]
1.534 [19]
1.230 [19]
1.812 [19]
1.507 [19]
1.450 [15]
2.209 [3]
3.222 [5]
2.531 [5]
2.864 [3]
2.929 [8]
3.053 [7]
1.862 [9]
1.547 [11]
2.832 [7]
2.200 [10]
1.938 [11]
2.71 [10]
2.353 [9]
4. Discussion
For describing the efficiency of proposed method Pearson correlation coefficients were calculated between sets of
indictors (JCR IF, WIF by H & Y, WIF by G. Buela-Casal, proposed WIF, average and median WIF, average and
median 5WIF) (Table 5). Results are illustrated graphically in Figures 1 to 15. Analyzing correlation coefficients among
different indicators for journal evaluation, we can notice that, proposed in this paper WIF have high correlations with all
analyzed indicators, excluding WIF proposed by G. Buela-Casal, which has low correlations with all indicators in this
list. The WIF proposed by H&Y and WIF proposed by G. Buela-Casal have the worst correlations, because these two
indicators deny each other: WIF proposed by H&Y decreases by increasing of cited journal IF in previous year, but WIF
proposed by G. Buela-Casal increases in the same case. The highest correlations are observed between average and
median 5-year weighted impact factors.
Table 5. Pearson correlation coefficients between each pair of journal evaluation indicators.
Indicators
JCR IF
Proposed
WIF by H
&Y
WIF by G.
Buela-Casal
Average
WIF
JCR IF
1
0.7398
0.6120
0.6105
0.6256
Proposed
WIF
1
0.7384
0.6226
0.8060
WIF by H
&Y
1
0.2797
0.7414
WIF by
Buela-Casal
1
0.6210
Average
WIF
1
Median
WIF
Average
5WIF
Median
5WIF
Alguliyev et al.
Median WIF
Average
5WIF
Median
5WIF
9
0.6361
0.7218
0.6932
0.7233
0.8135
0.7699
0.6045
0.8075
0.7473
0.6421
0.5774
0.5790
0.8887
0.9428
0.9489
1
0.8647
0.8948
1
1
0.9474
0.9474
1
1
5. Conclusion
25
20
15
10
5
0
WIF by H&Y
Proposed WIF
As the special case from equation of proposed weighted impact factor we get current equation of impact factor
calculated by ISI for JCR, but considering prestige and influence of cited journals gives more accurate indicator for
journal ranking. Suggested method for IF can also apply to researcher, organization or country evaluation. Obviously,
the paper or researcher cited by e.g. L. Zadeh is more important than object cited by unknown researcher. Therefore,
regarding indexes of cited objects for research evaluation contributes more effective results.
0
10
20
25
20
15
10
5
0
30
0
10
20
30
JCR IF
JCR IF
Figure 2. JCR IF versus WIF by H&Y
25
25
20
20
Average WIF
WIF by G. Buela-Casal
Figure 1. JCR IF versus proposed WIF
15
10
5
15
10
5
0
0
0
5
10
15
20
JCR IF
Figure 3. JCR IF versus WIF by G. Buela-Casal
25
0
5
10
15
20
JCR IF
Figure 4. JCR IF versus Average WIF
25
10
25
25
20
20
Average 5WIF
Median WIF
Alguliyev et al.
15
10
5
0
0
5
10
15
20
15
10
5
0
25
0
10
JCR IF
25
20
20
15
15
WIF by H&Y
Median 5WIF
Figure 6. JCR IF versus Average 5WIF
25
10
5
0
10
20
10
5
0
0
30
5
25
25
20
20
15
10
5
0
10
15
20
25
Figure 8. Proposed WIF versus WIF by H&Y
Average WIF
WIF by Buela-Casal
Figure 7. JCR IF versus Median 5WIF
5
10
Proposed WIF
JCR IF
0
30
JCR IF
Figure 5. JCR IF versus Median WIF
0
20
15
20
25
Proposed WIF
Figure 9. Proposed WIF versus WIF by Buela–Casal
15
10
5
0
0
5
10
15
20
Proposed WIF
Figure 10. Proposed WIF versus Average WIF
25
Alguliyev et al.
11
25
Average 5WIF
25
Median WIF
20
15
10
5
20
15
10
5
0
0
0
5
10
15
20
25
0
5
Proposed WIF
15
10
5
0
15
20
25
WIF by G. BUela-Casal
Median 5WIF
20
10
25
25
20
15
10
5
0
0
5
10
15
20
25
WIF by H&Y
Proposed WIF
Figure 13. Proposed WIF versus Median 5WIF
Figure 14. WIF by H&Y versus WIF by G. Buela-Casal
25
25
20
15
10
5
0
20
Median WIF
Average WIF
20
Figure 12. Proposed WIF versus Average 5WIF
25
5
15
Proposed WIF
Figure 11. Proposed WIF versus Median WIF
0
10
0
5
10
15
20
WIF by H&Y
Figure 15. WIF by H&Y versus Average WIF
25
15
10
5
0
0
5
10
15
20
WIF by H&Y
Figure 16. WIF by H&Y versus Median WIF
25
12
25
25
20
20
Median 5WIF
Average 5WIF
Alguliyev et al.
15
10
5
0
0
5
10
15
20
15
10
5
0
0
25
5
Figure 17. WIF by H&Y versus Average 5WIF
20
25
Figure 18. WIF by H&Y versus Median 5WIF
25
25
20
20
Median WIF
Average WIF
15
WIF by H&Y
WIF by H&Y
15
10
5
0
15
10
5
0
0
5
10
15
20
25
0
5
WIF by G. Buela-Casal
20
20
Median 5WIF
25
15
10
5
0
5
10
15
15
20
25
Figure 20. WIF by G. Buela-Casal versus Median WIF
25
0
10
WIF by G. Buela-Casal
Figure 19. WIF by G. Buela-Casal versus Average WIF
Average 5WIF
10
20
25
WIF by G. Buela-Casal
Figure 21. WIF by G. Buela-Casal versus Average 5WIF
15
10
5
0
0
5
10
15
20
25
WIF by G. Buela-Casal
Figure 22. WIF by G. Buela-Casal versus Median 5WIF
13
25
25
20
20
Average 5WIF
Median WIF
Alguliyev et al.
15
10
5
0
0
5
10
15
20
15
10
5
0
25
0
5
Average WIF
20
20
Average 5WIF
Median 5WIF
25
15
10
5
0
10
15
20
10
5
0
25
0
5
20
Median 5WIF
Median 5WIF
25
20
15
10
5
0
15
15
20
25
Figure 26. Median WIF versus Average 5WIF
25
10
10
Median WIF
Figure 25. Average WIF versus Median 5WIF
5
25
15
Average WIF
0
20
Figure 24. Average WIF versus Average 5WIF
25
5
15
Average WIF
Figure 23. Average WIF versus Median WIF
0
10
20
Median WIF
Figure 27. Median WIF versus Median 5WIF
25
15
10
5
0
0
5
10
15
20
25
Average 5WIF
Figure 28. Average 5WIF versus Median 5WIF
References
[1]
[2]
[3]
[4]
[5]
Garfield E. “Science citation index” – A new dimension in indexing. Science 1964; 144 (3619): 649-654.
Opthof T. Sense and nonsense about the impact factor. Cardiovascular Research 1997; 33 (1): 1-7.
Seglen PO. Citations and journal impact factors: Questionable indicators of research quality. Allergy: European Journal of
Allergy and Clinical Immunology 1997; 52(11): 1050-1056.
Reedjik J. Sense and nonsense of science citation analyses: comments on the monopoly position of ISI and citation
inaccuracies. New Journal of Chemistry 1998; 22: 767-770.
Gonzales-Pereira B, Guerrero-Bote VP, Moya-Anegon F. A new approach to the metric of journals’ scientific prestige: The
SJR indicator. Journal of Informetrics 2010; 4:379-391.
Alguliyev et al.
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
14
Perez-Hornero P, Arias-Nicolas JP, Pulgarin AA. An annual JCR impact factor calculation based on Bayesian credibility
formulas. Journal of Informetrics 2013; 7: 1-9.
Rousseau R. On the relation between the WoS impact factor, the Eigenfactor, the SCImago Journal Rank, the Article Influence
Score and the journal h-index. Preprint, http://hdl.handle.net/10760/13304.
Palacios-Huerta I, Volij O. The measurement of intellectual influence. Econometrica 2004; 72 (3): 963-977.
Bergstrom C. Eigenfactor: Measuring the value and prestige of scholarly journals. College & Research Libraries News 2007;
68 (5): 314-316.
Bollen J, Van de Sompel H, Hagberg A, Chute R. A principal component analysis of 39 scientific impact. Plos One 2009; 4
(6): e6022. doi:10.1371/journal.pone.0006022.
Lillquist E, Green S. The discipline dependence of citation statistics. Scientometrics 2010; 84 (3): 749-762.
Owlia P, Vasei M, Goliaei B, Nassiri I. Normalized Impact Factor (NIF): An adjust method for calculating the citation rate of
biomedical journals. Journal of Biomedical Informatics 2011; 44: 216-220.
Iftikhar M, Masood S., Song TT. Modified Impact Factor (MIF) at Specialty Level: A Way Forward. Procedia - Social and
Behavioral Sciences 2012; 69: 631-640.
Pudovkin A., Garfield E. Rank-normalized impact factor: a way to compare journal performance across subject categories.
Proceedings of the American Society for Information Science and Technology 2005; 41 (1): 507-515.
Leydesdorff L, Opthof,T. Scopus' SNIP Indicator. Preprint 2010; http://arxiv.org/abs/1005.4906
Pinski G, Narin F. Citation influence for journal aggregates of scientific publications: Theory, with application to literature of
physics. Information Processing and Management 1976; 12 (5): 297-312.
Bollen J., Rodriguez MA, Van de Sompel H. Journal status. Scientometrics 2006; 69 (3): 669-687.
Zyezkowski K. Citation graph, weighted impact factor and performance indices. Scientometrics 2010; 85: 301-315.
Buela-Casal G. Assessing the quality of articles and scientific journals: proposal for weighted impact factor and a quality
index. Psychology in Spain 2004; 8 (1): 60-76.
Habibzadeh F, Yadollahie M. Journal weighted impact factor: A proposal. Journal of Informetrics 2008; 2: 164-172.
Waltman L, Van Eck NJ. Some comments on the journal weighted impact factor proposed by Habibzadeh and Yadollahie.
Journal of Informetrics 2008; 2: 369-372.