Duplicate and fake publications in the scientific

literature: how many SCIgen papers in computer

science?
Cyril Labbe, Dominique Labbe

To cite this version:

Cyril Labbe, Dominique Labbe. Duplicate and fake publications in the scientific literature:
how many SCIgen papers in computer science?. Scientometrics, Akademiai Kiado, 2012,
pp.10.1007/s11192-012-0781-y. <10.1007/s11192-012-0781-y>. <hal-00641906v2>

HAL Id: hal-00641906

https://hal.archives-ouvertes.fr/hal-00641906v2
Submitted on 2 Jul 2012

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.

Larchive ouverte pluridisciplinaire HAL, est

destinee au depot et `a la diusion de documents
scientifiques de niveau recherche, publies ou non,
emanant des etablissements denseignement et de
recherche francais ou etrangers, des laboratoires
publics ou prives.

Duplicate and Fake Publications in the Scientific Literature: How

many SCIgen papers in Computer Science?
Cyril Labbe
Universite Joseph Fourier
Laboratoire dInformatique de Grenoble
First.Last@imag.fr

Dominique Labbe
Institut dEtudes Politiques de Grenoble
PACTE
First.Last@iep-grenoble.fr

22 june 2012 ; Scientometrics; DOI 10.1007/s11192-012-0781-y

Abstract
Two kinds of bibliographic tools are used to retrieve scientific publications and make them
available online. For one kind, access is free as they store information made publicly available
online. For the other kind, access fees are required as they are compiled on information
provided by the major publishers of scientific literature. The former can easily be interfered
with, but it is generally assumed that the latter guarantee the integrity of the data they sell.
Unfortunately, duplicate and fake publications are appearing in scientific conferences and, as
a result, in the bibliographic services. We demonstrate a software method of detecting these
duplicate and fake publications. Both the free services (such as Google Scholar and DBLP)
and the charged-for services (such as IEEE Xplore) accept and index these publications.

keyword: Bibliographic Tools, Scientific Conferences, Fake Publications, Text-Mining, InterTextual Distance, Google Scholar, Scopus, WoK

Introduction

Several factors are substantially changing the way the scientific community shares its knowledge. On the one hand, technological developments have made the writing, publication and
dissemination of documents quicker and easier. On the other hand, the pressure of individual evaluation of researcherspublish or perishis changing the publication process. This
combination of factors has led to a rapid increase in scientific document production. The three
largest tools referencing scientific texts are: Scopus (Elsevier), ISI-Web of Knowledge (WoK
Thomson-Reuters) and Google Scholar.
Google Scholar is undoubtedly the tool which references the most material. It is free and
it oers wide coverage, both of which are extremely useful to the scientific community. Google
Scholar allows grey literature to be more visible and more accessible (technical reports, long versions and/or tracts of previously published papers, etc). Google Scholar systematically indexes
everything that looks like a scientific publication on the internet, and, inside these documents
and records, it indexes references to other documents. Thus, it gives a picture of which documents are the most popular. However, the tool, much like the search engine Google, is sensitive
to Spam [2], mainly through techniques, similar to link farms that artificially increase the
ranking of web pages. Faked papers like those by Ike Antkare [12] (see 2.2 below) may also
be mistakenly indexed. This means that documents indexed by Google Scholar are not all bona
fide scientific ones, and information on real documents (such as the number of citations found)

can be manipulated. This type of tool, using information publicly and freely available on the
Web, faces some reproducibility and quality control problems [22, 10].
In comparison, editorial tools (such as Scopus or WoK) seem immune to this reproach.
They are smaller, less complete and require access fees, but in return they may be considered
as cleaner. This is mainly because they store only publications in journals and conferences
in which peer selection is supposed to guarantee the quality of the indexed publications. The
number of citations is computed in a more parsimonious way and meets more stringent criteria.
Data quality would also seem to be secured by a new selection by the publisher who provide
the tool:
This careful process helps Thomson Scientific remove irrelevant information and present
researchers with only the most influential scholarly resources. A team of editorial experts,
thoroughly familiar with the disciplines covered, review and assess each publication against
these rigorous selection standards[11]1 .
Dierences between these tools have been studied [7, 25, 9]. But are they immune from
failures such as multiple indexing of similar or identical papers (duplicates), or even the indexing
of meaningless publications?
A first answer to these questions will be provided by the means of several experiments on
sets (corpora) of recent texts in the field of Computer Science. Text-mining tools are presented
and used to detect problematic or questionable papers such as duplicated or meaningless publications. The method has enabled the identification of several bogus scientific papers in the
field of Computer Science.

Corpora and texts preprocessing

Table 1 gives a synthetic view of the sets of texts used along this article2 .
A priori above-reproach corpora: Most of the texts used in these corpora are indexed in
bibliographic tools (Scopus and WoK). They are either available from the conferences web sites,
or from the publishers web sites, like the Institute of Electrical and Electronic Engineers (IEEE)
or Association for Computing Machinery (ACM) websites, which sponsor a large number of
scientific events in the field of electronics and computer science. Acceptance rates are published
by the conferences chairs in the proceedings. Texts of corpora X, Y and Z were published in
three conferences (X, Y and Z). The MLT corpus is composed of texts published in various
conferences. They have been retrieved by applying, to 3 texts of the corpus Y, the More Like
This functionality provided by IEEE (see figure 1).
Representative set of articles in the field of Computer Science: ArXiv is an open
repository for scholarly papers in specific scientific fields. It is moderated via an endorsement
system which is not a peer review: We dont expect you to read the paper in detail, or verify
that the work is correct, but you should check that the paper is appropriate for the subject
area3 .
All the computer science papers for the years 2008, 2009 and 2010 were downloaded from
the arXiv repository. Excluding the ones from which text could not be extracted properly this
represent: 3481 articles for year 2008, 4617 for 2009 and 7240 for 2010.
1

http://ip-science.thomsonreuters.com/news/2005-04/8272986/
Bibliographic information and corpora are available upon request to the authors
3
http://arxiv.org/help/endorsement
2

Table 1: Corpora description: NA stand for non available.

Corpus
name

Downloaded
from

Corpus X

ACM
portal.acm.org

Corpus Y

Corpus Z

Years

Type
of papers

Number
of papers

Acceptance
rate

Corpus
size

2010

Full
Short
Demo

126
165
20

13.3%
17.5%
52%

311

2009

Regular

150

28%

150

Track 1
Track 2
Track 3
Demo

58
33
36
32

18.4%
16.1%

153

200x-20yy

various

122

NA

122

2008
2009
2010

various

3481
4617
7240

NA
NA

15338

IEEE
ieee.org
Conf.
Web Site

MLT

IEEE
ieee.org

arXiv

arxiv.org

2010

36%

Figure 1: The More Like This functionality was applied to 3 texts of the Y corpus.

Automatically generated, deliberately faked texts: These corpora contain documents

automatically generated using the software SCIgen4 . This software, developed at MIT in 2005,
generates random texts without any meaning, but having the appearance of research papers
in the field of computer science, and containing summary, keywords, tables, graphs, figures,
citations and bibliography. Table 2 shows the first words for some of the 13 possible sentences that start a SCIgen paper. Inside these sentences, token starting with SCI are randomly
chosen among predefined words. For example, SCI PEOPLE have 23 possible values including:
steganographers, cyberinformaticians, futurists or cyberneticists. SCI BUZZWORD ADJ have 74
possible values such as: omniscient, introspective, peer-to-peer or ambimorphic. The whole
SCIgen grammar have almost four thousand lines and is fairly complex. Texts are also embellished with rather eccentrics graphs and figures. This allows the generation of a very large set
of dierent texts syntactically correct but without any meaning, which can be spotted quite
easily.
4

http://pdos.csail.mit.edu/scigen/

Table 2: First words of sentences that start a SCIgen-Origin paper.

Many SCI PEOPLE would agree that, had it not been for SCI GENERIC NOUN , ...
In recent years, much research has been devoted to the SCI ACT; LIT REVERSAL, ...
SCI THING MOD and SCI THING MOD, while SCI ADJ in theory, have not until ...
The SCI ACT is a SCI ADJ SCI PROBLEM.
The SCI ACT has SCI VERBED SCI THING MOD, and current trends suggest that ...
Many SCI PEOPLE would agree that, had it not been for SCI THING, ...
The implications of SCI BUZZWORD ADJ SCI BUZZWORD NOUN have ...

For the Antkare experiment, SCIgen was modified so that each article had references to
the 99 otherscreating a link farm. Thus, all these texts have the same bibliography. Google
Scholar retrieved these faked online articles and, as a result, Ike Antkares H-index reached 99,
ranking him in the 21st position of the most highly cited scientists [12].
The corpus Antkare is composed of the 100 documents used for this experiment. 236 articles
generated by the original version of the SCIgen software compose the corpus SCIgen-Origin.
At least one other version of SCIgen exists. It is an adaptation of the original SCIgen for
physics, especially solid state physics and neutron scattering5 . A set of 414 articles generated
by this software will be referred in the following as the corpus SCIgen-Physics.
Table 3: SCIgen Corpora
Corpus name

Generator

Scientific field

Corpus size

SCIgen-Origin

Original SCIgen

Computer Science

236

Antkare

Modified SCIgen

Computer Science

100

SCIgen-Physics

Modified SCIgen

Physics

414

Table 3 gives a synthetic view of the used SCIgen corpora, examples of SCIgen-Origin and
SCIgen-Physics can be found in appendix A.
Texts Processing: Pdf files are converted to plain text files by the program pdftotxt (free
software unix and windows version 3.01) that extracts the text from pdf files. During this
operation, figures, graphs and formulas disappear, but the titles and captions of these figures
and tables remain. To prevent the 100 identical references in the corpus Antkare from disturbing
the experiments, the bibliographies (and appendices) have been removed from all texts in all
corpora.
The texts are segmented into word-tokens using the Oxford Concordance Program commonly
used for English texts [8]. In fact, the word-tokens are caracter strings separated by spaces or
punctuation. This procedure could be further improved for example by word tagging to replace
all the abbreviations and inflections of a single word with a unique spelling convention (infinitive
form of verbs, singular masculine of adjectives, etc.)
5
Blog post: http://pythonic.pocoo.org/2009/1/28/fun-with-scigen
SCIgen-Physics Sources: https://bitbucket.org/birkenfeld/scigen-physics/overview

Text mining tools

Distances between a text and others (inter-textual distances) are computed. Then these distances are used to determine which texts, within a large set, are closer to each other and may
thus be grouped together.
Inter-textual distance: The distance between two texts A and B is measured using the
following method (previous work in [13, 14]). Given two texts A and B, let us consider:
NA and NB : the number of word-tokens in A and B respectively, ie the lengths of these
texts;
FiA and FiB : the absolute frequencies of a type i in texts A and B respectively;
|FiA FiB | the absolute dierence between the frequencies of a type i in A and B respectively;
D(A,B) : the inter-textual distance between A and B is as follows:
D(A,B) =

|FiA FiB |

with NA = NB

(1)

i(AB)

The distance index (or relative distance) is as follows:

Drel(A,B) =

"

i(AB) |FiA

FiB |

NA + NB

(2)

This index can be interpreted as the proportion of dierent words in both texts. A distance
of 0.4 means that the texts share 60% of their words-token.
If the two texts are not of the same lengths in tokens (NA < NB ), B is reduced to the
length of A:
U=

NA
NB

is the proportion used to reduce B in B

EiA(u) = FiB .U is the theoretical frequency of a type i in B

In the Equation (1), the absolute frequency of each word-type in B is replaced by its theoretical frequency in B :
!
|FiA EiA(u) |
D(A,B ) =
i(AB)

Putting aside rounding-os, the sum of these theoretical frequencies is equal to the length
of A. The Equation (2) becomes:
Drel(A,B) =

"

i(AB) |FiA

EiA(u) |
NA + NB

This index varies evenly between 0 the same vocabulary is used in both texts (with the
same frequencies) and 1 (both texts share no word-token). An inter-textual distance of
can be interpreted as follows: choosing randomly 100 words in each text, is the expected
proportion of common words between this two sets of 100 words.
In order to make this measure fully interpretable:
the texts must be long enough (at least more than 1000 word-tokens),
5

 one must consider that, for short texts (less than 3000 word-tokens), values of the index
can be artificially high and sensitive to the length of the texts, and
the lengths of the compared texts should not be too dierent. In any case, the ratio of
the smallest to the longest must be less than 0.1.
Inter-textual distance depends on four factors. In order of decreasing importance, they are
as follows: genre, author, subject and epoch. In the corpora presented above, all texts are in
the same genre (scientific papers) and are contemporary. Thus only the authorial and thematic
factors remain to explain some anomalies.An unusually small inter-textual distance suggests
striking similarities and/or texts by the same author.
Agglomerative Hierarchical Clustering: The inter-textual distances allow agglomerative
hierarchical clustering according to similarities between texts and graphical representations of
their proximities [23, 3, 20, 21].
This representation is used to identify more or less homogeneous groups in a large population.
The best classification is the one that minimizes the distances between texts of the same group
and maximizes the distances between groups.
An agglomerative hierarchical clustering is performed on the inter-textual distance matrix,
using the following method. The algorithm proceeds by grouping the two texts separated by
the smallest distance and by recomputing the average (arithmetic mean) distance between all
other texts and this new set, and so on until the establishment of a single set.
These successive groupings are represented by a dendrogram with a scale representing the
relative distances corresponding to the dierent levels of aggregation (see Figure 3 and 4).
By cutting the graph, as close as possible to a thresholds considered as significant, one can
demarcate groups of texts as very close, fairly close, etc. The higher the cut is made, the more
heterogeneous the classes are and the more complex is the interpretation of the dierences. To
correctly analyze these figures, it must be also remembered that:
whatever their position on the non-scaled axis, the proximity between two texts or groups
of texts is measured by the height at which the vertices uniting them converge, and
the technique sometimes results in chain eects: some similarities between texts are indistinguishable because the vertices connecting them are erased by aggregations performed
at a lower level.
Related work: One can find, in the scientific literature, several indices for measuring the
similarities (or dissimilarities) between texts. Most often, these indices are based on the vocabulary matrix. Cosine and Jaccard indexes are frequently used and they seem to be well adapted
to texts [16]. Some indices based on compression have also been tested [17]. Compared to
these indices, intertextual distance is easily interpretable: it is a measure of the proportion of
word-tokens shared by two texts. Based on frequencies it could be interpreted as being closely
related to information theory: having always the same word-types at the same frequencies do
not provide any new information.
In the past recent years, some methods have been developed aiming at automatically identifying SCIgen papers. [24] checks whether references are proper references that points to
documents known by the databases available online. A paper having a large proportion of
unidentified references will be suspected to be a SCIgen paper. An other approach is proposed
in [15]. This method is based on an ad-hoc similarity measure in which the reference section
plays a major role. These characteristics explain why these techniques were not able to identify
6

texts by Ike Antkare as being SCIgen paper6 . A third proposition [5] is based on observed
compression factor and a classifier. A paper under test will be classified as being generated if
it has a compression factor similar to known generated text. The method focuses on detecting
SCIgen paper but also, what is more, on detecting any kind of texts generated automatically7 .
A simple test shows that this software wrongly classifies as authentic the texts by Antkare
(when their reference sections are not withdrawn), with around 10% risks of error, and that it
identifies the same texts as inauthentic, when their reference sections are withdrawn... Finally,
again, these methods do not provide an easily interpretable procedure for the comparison of
texts (in contrast with intertextual distance).
Interesting questions: Like most of the metrics of textual similarities, inter-textual distance, is based on the so called bag-of-word approach. Such measures are sensitive to word
frequencies but insensitive to syntax. Using this kind of approach to detect SCIgen papers relies
on the fact that, despite its wide range of preset sentences, the SCIgen vocabulary remain quite
poor: SCIgen is behaving like an author that would have been poorly gifted with vocabulary.
The combination of intertextual distance with agglomerative hierarchical clustering allows
some interesting questions to be answered. For example, do the conferences under consideration
contain the following occurencies?
chimeras comparable to the texts by Ike Antkare
duplicates: the same authors present the same text twice under dierent titles
related papers: covering a wide range of cases, going from almost unchanged texts to
close texts by the same author(s) dealing with the same topics, sometimes sharing similar
portions of text. The scientific contents of these texts may be substantially dierent. The
proposed tools do not provide any help to measure these dierences.

Detection of forgeries, duplicates and related papers in the

three conferences X, Y and Z

Intra-corpus distances: For each corpus, distances are ranked by ascending values and
distributed in equal interval classes. Fig. 2 shows these distributions.
The X, Y and Z corpora have the classic bell curve profile suggesting the existence of
relatively homogeneous populations (here a large number of contemporary authors writing in a
similar genre and on more or less similar themes). X and Z have a comparable mean/mode and
a similar dispersion. In contrast,
Y has a high average distance and a higher dispersion around this mean, indicating heterogeneity of papers, but also suggesting the presence of anomalies (these two explanations
are not mutually exclusive);
On the left of the graph, the curve with three modes is the distribution of distances between
the 100 faked texts by Ike Antkare. This trimodal distribution suggests the existence of
two dierent populations within the texts generated by the modified SCIgen: a small
group with very low internal distances are centered on 0.2 - these are short texts (about
1600 word-tokens) - and the other group, with a greater number of texts, containing longer
texts (about 3000 word-tokens): Their internal distances are centered on 0.38. The third
mode is distances between these two groups.
6
7

http://paperdetection.blogspot.com/
http://montana.informatics.indiana.edu/cgi-bin/fsi/fsi.cgi

12

Antkare
10

X
Z

6
0

Frequency

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Distance

Figure 2: Distribution of intra-corpus distances.

Main Groups: The classification and its representation by a dendrogram (Figure 3) show
four main groups:
In the center, a large body (C) includes all texts Z and almost all X texts. It would be
possible to isolate various subgroups within this group to show what are the main topical
themes of these conferences.
on the right (D) and on the extreme left (A), the texts of the Y conference meet at the
higher levels, confirming the heterogeneity of this conference.
There is very little intermingling between X, Z on one side and Y on the other side: only
six Y papers are included into X-Z set, but they are attached, at a very high level, to this
set (i.e. with significant distances). Similarly, only four X papers are included in group
A (Y). In other words, most of the papers presented at the Y conference are not of the
same nature as those presented at the other two conferences.
Finally, all the chimeras generated by SCIgen for Ike Antkare are grouped in B into two
homogeneous groups and connected at a very low level. Thus, SCIgen texts are not close to
natural language and are distinct from the scientific papers they are supposed to emulate.
Four genuine-fake texts: In the dendrogram in Figure 3, the number (1) branches are
four Y texts that are clustered within the corpus Antkare.
These four texts are genuine publications because they have, at least formally, been selected by peer reviewers. They are real publications also because they are in conference

l l
l l lll l
l
l
l
8
l l l l l
l
l
l
l
l
l
l lll
l
lll ll l l l
l
l
l
l l
l
l
ll l l l l lll
l ll
l
l ll | |
| |I |
|
| |
|| | |||I
|
7
| | | | || |
|
|
|
| || || |
| ||| | | | |
| |
|||| | || | |I
|| || | |
| | || |
|| | |
|
|
|
|
||
| | |I |||| |
6
| || || |
| ||I |
I
|
|
I
|
| |I
I
5
|I| | | I I I
I
I
| II | I
I
|I| |I|I
4
|II II | I| |
I
I
I
I
|
|I
I
I
III
I|
|II|I II
I
I I I| II II II
I I
I I | I
| I I| II I I
II I II I I I
I
I I
| I I |I |
||| ||| | | |
| I| |I|| |
|
| |
I | | I| I| | I
I
I
I
I | I I||I I
II|| |I
|
I| | | I I
|| | | |
3
||| || || |
| || I| | ||
|| | |
I|
|l
| I| | I I| I
| |II
l | || |l
|
|
|
|
|
I
|l | |l |l l
|
! !! !! !
!!!!
!!
!!! !! !
!
!
! !!!!! !!
!
!
!! ! !!
!! !
! !!! ! l l 1
l

l| | ll l| l l
l l
l l ll ll l l
l
l
0.7

0.6

B
!

ll

!!!!!
!!!!!!!!
!! !!!

l
0.5

0.4

0.3

0.2

0.1

0.0

Distance scale
Y

Antkare

Figure 3: Dendrogram for cluster analysis of corpora Antkare (black), X (green), Z (blue), Y
(red). Main clusters: group A (corpus Y), group B (corpus Ike Antkare), group C (corpora Z
et X), group D (corpus Y). Main remarkable points: (1) four Y texts are classified with Ike
Antkare fake documents, (2) two Y texts with a quasi zero distance, (3) and (7) two X texts
with a small distance, (4) and (5) are two couples of Z texts with a small distance, (6) a Z text
and a X text with a small distance, (8) two Y texts with a small distance,
9

proceedings. At the very least, because they are available (on payment) and referenced by sites
of serious and professional scientific publishers (Web of Science, Scopus, IEEE).
But these texts are fake publications because they have the characteristics of the texts
generated using SCIgen: absurd titles and figures, faked bibliographies, mixture of jargon with
no logic.
Duplicates publications: Number (2) branch is a zero distance (0.006) between two Y
papers. Only the titles are dierent. It reveals that an identical text have been published twice,
the same year in the same conference.
Smallest distances (without SCIgen texts): The branches of the dendrogram numbered
(3) to (8) are the texts with the smallest distances all sharing a common subset of authors and
very similar topics. They may be seen as related papers published the same years in the same
conference (or two dierent ones for branch (6)).

How many pseudo publications are in the online computer

science literature?

Answering this question would require a scan of the entire recently published literature in the
field of computer science. We consider here a more restricted question: Are the 4 pseudo texts
of the Y Conference unique? We will respond with a trial in the IEEE and arXiv databases.
A trial: The IEEE search engine oers a functionality (More Like This in figure 1) that
researches texts, similar to a chosen paper. We applied it to three SCIgen papers from Y
corpus. On the day of the experiment (April 22, 2011), this functionality returned 122 dierent
documents that, therefore, the IEEE considers to be close to these SCIgen papers. We call this
new corpus More Like This MLT and we applied to it the same tools. To make this cluster
analysis readable, the dendrogram, reproduced in Figure 4, relates only the comparison of this
new corpus with the Antkare texts (to detect some new SCIgen texts) and with those of Z
(containing only genuine texts).
It appears that the corpus MLT includes:
81 new pseudo papers grouped with Ike Antkare documents (Group C Figure 4). C1
contains 17 texts very similar to those of Ike Antkare, but slightly distorted to pass the
peer selection. Careful examination of these papers shows that sometimes the titles are
appropriate to the subject of the conference, some abstracts are more or less coherent,
and few figures have been changed, but most of the writing remains SCIgen. C2 contains
64 twins from those of Ike Antkare. Careful reading of these texts reveals that the texts
generated by SCIgen were published, without any change. C3 and C4: twice, identical
SCIgen papers were presented under dierent titles, by the same authors to two dierent
conferences.
41 genuine papers are classified into two groups (A and B).
Careful reading reveals that some of these 41 texts are not above suspicion (especially for
the group A in Figure 4). Several passages contain inconsistent text or texts unrelated to the
rest, one bibliography, at least, comes from SCIgen. But all these articles are clearly not SCIgen
Computer Science generated texts.
The cluster analysis shows 14 quasi-duplicate or related papers, which correspond to five
groups A1, A2 and A3, B1 and B2.
10

A1
l
l l

l l

Distance

0.6

l
l
l I I
II I
I I I II
I II
III
I II
I
I
I II
I II
I
I II I
I
I
I
III I
I
I II I I
I
II II
I II I I
I
I
II I I I I
I
I
I
I
II I
I
II
II
Il I I
II
I I
l
I
I II I
I
I
I
I
II I
l l
Il
I
I
I I
I
l l
I
l
l l
ll l
l
l
l
lll l
|||
|| |
|
|
l
||||
|| || |
||| |
||| | l
l| |
l l|| l
| l |l
l l ll
l|
l l l| l
|
| | ||
|| |
l l ll
l l ll
ll l l
l lll l
l
| | lll|
l|
|
l
l l|

0.5

0.4

0.3

A3

B1

B2

C1

C2

C3
l

C4
|
|| ||
|||
||||
|
| ||||
||

0.2

0.7

A2

0.1

0.0

Antkare

Figure 4: Dendrogram for analysis of corpora Antkare (black), Z (blue), MLT (red). Main
clusters: C (Antkare and MLT SCIgen texts), B (Z and MLT genuine), A (MLT genuine).
Main remarkable points: C3 C4 (pseudo papers published twice). A1, A2, A3, B1, B2 (related
papers).
11

In one case, both documents correspond to the same paper at dierent stages. First presented in a conference, the paper was then deemed worth being published, with some modifications, in a scientific journal. Of course, these two documents should be indexed together. In
this case, it is simple since the authors and the titles are the same. If search engines could be
able to detect this kind of frequent occurrence, this could provide a fruitful help to users.
Automatic detection of SCIgen papers: A nearest neighbor classification (knn classification [4, 18] with k=1) was tested to verify the feasibility of automatic detection of pseudo
papers. For this experiment, the 100 documents of the Ike Antkare corpus and the 121 articles
of the Z corpus respectively represent the fake and genuine papers. A 1-nn classification
is done to assign each MLT article to the class of its nearest neighbor. So, for each text of the
corpus More Like This the distances to the 221 reference texts are computed and the text is
assigned to the group of its nearest neighbor.
Using this method all pseudo items (group C in figure 4) are classified with the corpus
Antkare. Observed distances to the closest neighbor in the Corpus Antkare are ranging from
0.33 to 0.52. Detailed reading of the paper with this 0.52 distance reveals that it contains at
least 30% of SCIgen computer science generated text. Some other parts of the paper seams also
directly adapted from SCIgen. Its distance to its closest neighbor in the set of genuine paper
of the Z corpus is 0.56 which suggest its alien status.
Risk of misclassifying SCIgen papers: Is there a risk of misclassifying a SCIgen paper as a
genuine one? This risk is assessed thanks to the two corpora SCIgen-Origin and SCIgen-Physics.
All the 236 SCIgen-Origin texts are well classified as being generated papers. Distances to their
closest neighbors in the Corpus Antkare range from 0.32 to 0.37. All the 414 SCIgen-Physics
articles are also well classified in the Corpus Antkare. For this last corpora, distances to the
closest neighbors in the Corpus Antkare are ranging from 0.42 to 0.48.
These results show that the proposed method should hardly misclassify a SCIgen paper as
being a non-SCIgen one.
Risk of misclassifying non-SCIgen papers: Is there a risk of misclassifying a genuine
paper as being generated by SCIgen? The arXiv corpus is used to evaluate this risk. Out of
the arXiv Corpus, eight texts are classified with SCIgen papers with distances to their nearest
neighbors in the Corpus Antkare greater than 0.9: these eight texts are not written in English.
Only one English paper was wrongly classified as being a SCIgen paper. Its distance to its
closest neighbor in the Antkare Corpus is 0.621 to be compared to its closest neighbor in the Z
corpus 0.632. Such distances should suggest that this text, and the SCIgen ones, are not of the
same kind.
Following this standard classification process the risk of misclassifying a genuine document
as being SCIgen can be estimated to 1/15000 = 6.5 105 . A simple way to avoid this kind
of false positive is to adopt the following rule: a text under test should not be classified as
being SCIgen if its distance, to its nearest neighbor in the fake corpora, is greater than a
threshold. Given the previously exposed experiments (MLT Corpus), this threshold could be
set around 0.55. Over such a distance, no conclusion can be drawn out. Under this threshold,
the hypothesis of a SCIgen origin must be seriously considered. This last method has been
adopted to provide a web site oering SCIgen detection8 .
8

http://sigma.imag.fr/labbe/main.php

12

Conclusions

Scope of the problem? In total, the 85 SCIgen papers identified have the following characteristics:
89 dierent authors, 63 of whom have signed only one pseudo publication. In contrast,
three have signed respectively 8, 6 and 5. These three authors belong to the same
university;
These 89 authors belong to 16 dierent universities. One such university is the origin
of a quarter of these 85 pseudo papers;
24 dierent conferences have been infected between 2008 and 2011. For the most aected
there was 24 and 11 fake papers published.
It can be reasonably assume that, the reviewers, at least 85 times in 24 dierent conferences,
have missed completely meaningless papers, or the ones having been altered with a few cosmetic
improvements. Because these publications are then indexed in the bibliographic tools, these
repositories may include a certain number of anomalies. A large scale experiment would be
needed to estimate the number of duplicates, near-duplicates and fake papers in the IEEE
database which contains more than 3,000,000 documents. It may be a marginal or minor
problem, but the fee-based databases should cope with it better than the free ones.
On the other hand, on the days when arXiv documents were downloaded9 , none of them
were SCIgen generated (at least the one for which txt could be extracted).
Why these phenomena? As for the authors, the pressure of publish or perish may explain,
but not excuse, some anomalies. SCIgen software was designed to test some conferencesthe
selection process of which seemed dubiousproviding them with contrived bogus articles. But
the deception was announced and the chimera was withdrawn from the proceedings [1]. This,
however, is not the case for the 85 pseudo texts that we detected.
Since 2005, the number of international conferences has been increasing. Most of these
conferences cover a wide spectrum of topics (such as conference Y analyzed in this article).
This is their Achilles heel: Their reviewers may not be competent on all the topics announced
in the conference advertisements. Ignoring the jargon of many sub-disciplines, they may think:
I do not understand it, but it seems to be of depth and bright. A reflexion on how could a
good conference be characterized can be found in [6].
Textual data mining tools would be eective tools for analysis and computer-aided decisionmaking. The experiments suggest that they are of significant interest in detecting anomalies
and allowing conference organizers and managers of databases to eliminate them. The use of
such tools would also be an excellent safeguard against some malpractices.
Of course, automatic procedures are only an aid and not a substitute for reading. The
double-checking evaluation by attentive readers remains essential before any decision is made
to accept and publish. Similarly, in order to evaluate a researcher or a laboratory, the best way
is still to read their writings [19].
acknowledgements: The authors would like to thank Tom Merriam, Jacques Savoy, Edward
Arnold for their careful readings of previous versions of this paper, the anonymous reviewers
and members of the LIG laboratory for their valuable comments.
9

February and March 2012

13

References
[1] Ball, P.: Computer conference welcomes gobbledegook paper. Nature 434, 946 (2005)
[2] Beel, J., Gipp, B.: Academic search engine spam and google scholars resilience against it.
Journal of Electronic Publishing 13(3) (2010). URL http://hdl.handle.net/2027/spo.
3336451.0013.305
[3] Benzecri, J.P.: Lanalyse des donnees. Dunod (1980)
[4] Cover, T.M., Hart, P.E.: Nearest neighbor pattern classification. IEEE Transactions on
Information Theory 13, 2127 (1967)
[5] Dalkilic, M.M., Clark, W.T., Costello, J.C., Radivojac, P.: Using compression to identify
classes of inauthentic texts. In: Proceedings of the 2006 SIAM Conference on Data Mining
(2006)
[6] Elmacioglu, E., Lee, D.: Oracle, where shall i submit my papers? Communications of the
ACM (CACM) 52(2), 115118 (2009)
[7] Falagas, M.E., Pitsouni, E.I., Malietzis, G.A., Pappas, G.: Comparison of pubmed, scopus,
web of science, and google scholar: strengths and weaknesses. The FASEB Journal 22(2),
338342 (2008)
[8] Hockey, S., Martin, J.: OCP Users Manual. Oxford. Oxford University Computing Service
(1988)
[9] Jacso, P.: Testing the calculation of a realistic h-index in Google Scholar, Scopus, and Web
of Science for F. W. Lancaster. LIBRARY TRENDS 56(4) (2008)
[10] Jacso, P.: The pros and cons of computing the h-index using Google Scholar. Online
Information Review 32(3), 437452 (2008). DOI 10.1108/14684520810889718. URL http:
//dx.doi.org/10.1108/14684520810889718
[11] Kato, J.: Isi web of knowledge: Proven track record of high quality and value. KnowledgeLink newsletter from Thomson Scientific (April 2005)
[12] Labbe, C.: Ike antkare, one of the great stars in the scientific firmament. International
Society for Scientometrics and Informetrics Newsletter 6(2), 4852 (2010)
[13] Labbe, C., Labbe, D.: Inter-textual distance and authorship attribution corneille and
moliere. Journal of Quantitative Linguistics 8(3), 213231 (2001)
[14] Labbe, D.: Experiments on authorship attribution by intertextual distance in english.
Journal of Quantitative Linguistics 14(1), 3380 (2007)
[15] Lavoie, A., Krishnamoorthy, M.: Algorithmic Detection of Computer Generated Text.
ArXiv e-prints (2010)
[16] Lee, L.: Measures of distributional similarity. In: 37th Annual Meeting of the Association
for Computational Linguistics, pp. 2532 (1999)
[17] Li, M., Chen, X., Li, X., Ma, B., Vitanyi, P.: The similarity metric. Information Theory,
IEEE Transactions on 50(12), 32503264 (2004)
[18] Meyer, D., Hornik, K., Feinerer, I.: Text mining infrastructure in r 25(5), 569576 (2008)
14

[19] Parnas, D.L.: Stop the numbers game. Commun. ACM 50(11), 1921 (2007)
[20] Roux, M.: Algorithmes de classification. Masson (1985)
[21] Roux, M.: Classification des donnees denquete. Dunod (1994)
[22] Savoy, J.: Les resultats de google sont-ils biaises ? Le Temps (2006)
[23] Sneath, P., Sokal, R.: Numerical Taxonomy. San Francisco : Freeman (1973)
[24] Xiong, J., Huang, T.: An eective method to identify machine automatically generated
paper. In: Knowledge Engineering and Software Engineering, 2009. KESE 09. PacificAsia Conference on, pp. 101102 (2009)
[25] Yang, K., Meho, L.I.: Citation analysis: A comparison of google scholar, scopus, and web
of science. In: American Society for Information Science and Technology, vol. 43-1, pp.
115 (2006)

Examples of SCIgen papers.

Figure 5 is an example of a SCIgen-Physics paper. Formula generation have been improved

compare to the one used by SCIgen-Origin (cf figure 6).

15

Decoupling the Higgs Sector from Correlation in

Magnetic Scattering
A BSTRACT

0.25
0.2
free energy (dB)

Unified stable symmetry considerations have led to many

private advances, including tau-muons and hybridization [1].
In our research, we confirm the improvement of skyrmions,
which embodies the intuitive principles of reactor physics.
Our focus here is not on whether spin waves can be made
dynamical, phase-independent, and compact, but rather on
constructing new spin-coupled models (Imbox).

0.15
0.1
0.05
0
-0.05

I. I NTRODUCTION
Many chemists would agree that, had it not been for
spin-coupled Monte-Carlo simulations, the development of
correlation effects might never have occurred. Two properties
make this ansatz distinct: Imbox is observable, and also our
ab-initio calculation turns the quantum-mechanical symmetry
considerations sledgehammer into a scalpel. In this paper,
we argue the investigation of the Higgs boson. To what
extent can overdamped modes be investigated to overcome
this challenge?
Imbox, our new instrument for Bragg reflections with j < 53 ,
is the solution to all of these obstacles. Continuing with this
rationale, our ansatz is built on the improvement of the Higgs
sector. While conventional wisdom states that this quandary is
never overcame by the theoretical treatment of the positron, we
believe that a different approach is necessary. The flaw of this
type of method, however, is that tau-muon dispersion relations
with = 1 and the Fermi energy are generally incompatible.
Certainly, two properties make this method ideal: our approach
harnesses Landau theory, and also our instrument prevents
pseudorandom theories. This combination of properties has
not yet been harnessed in related work.
The rest of this paper is organized as follows. For starters,
we motivate the need for Einsteins field equations. Following
an ab-initio approach, we demonstrate the theoretical treatment
of excitations that would make controlling a gauge boson a
real possibility. Furthermore, we confirm the development of
electrons [1]. As a result, we conclude.
II. Imbox I MPROVEMENT
Imbox relies on the intuitive theory outlined in the recent
much-touted work by Eugene Wigner in the field of solid
state physics. Following an ab-initio approach, to elucidate
the nature of the electron dispersion relations, we compute
the electron given by [2]:
#
#

r ) = d3 r .
(
(1)
W

-0.1
-80 -60 -40 -20

20

40

60

Fig. 1.

The main characteristics of interactions.

We consider a theory consisting of n Einsteins field equations.

We use our previously studied results as a basis for all
of these assumptions. This follows from the estimation of
paramagnetism.
Our instrument is best described by the following relation:
$
%

k[]
= sin
(2)
,
n
where r is the rotation angle except at Z , we estimate broken
symmetries to be negligible, which justifies the use of Eq. 3.
we assume that particle-hole excitations and interactions can
connect to overcome this quandary [3], [4]. Figure 1 depicts
the schematic used by our model.
III. E XPERIMENTAL W ORK
As we will soon see, the goals of this section are manifold.
Our overall measurement seeks to prove three hypotheses:
(1) that the spectrometer of yesteryear actually exhibits better
free energy than todays instrumentation; (2) that a proton no
longer impacts system design; and finally (3) that average
free energy is even more important than a phenomenologic
approachs normalized count rate when improving integrated
electric field. Our analysis holds suprising results for patient
reader.
A. Experimental Setup
Though many elide important experimental details, we
provide them here in gory detail. We measured a time-of-flight
inelastic scattering on the FRM-II cold neutron diffractometers
to measure superconductive Monte-Carlo simulationss lack
of influence on the work of Italian theoretical physicist F.

Figure 5: Generated text, graph and formula : SCIgen Physics.

16

80 100

volume (mSv)

Decoupling Multicast Methods from Superblocks in

Robots

Abstract

metries. Contrarily, linear-time models might

not be the panacea that information theorists expected. Our heuristic prevents random technology. For example, many systems manage the evaluation of vacuum tubes.
However, this approach is never well-received.
Our mission here is to set the record straight.

The steganography solution to Internet QoS

is defined not only by the visualization of
RPCs, but also by the unfortunate need for
Markov models. Given the current status of
ecient algorithms, researchers predictably
desire the improvement of link-level acknowlWe confirm that the transistor and multiedgements, which embodies the important
cast frameworks are continuously incompatiprinciples of cryptography. HugyBoss, our
ble. This is often a private objective but has
new heuristic for telephony, is the solution
ample historical precedence. Contrarily, this
to all of these challenges.
approach is always considered robust. The
drawback of this type of approach, however,
is that Lamport clocks can be made secure,
1 Introduction
empathic, and cacheable. We emphasize that
our methodology improves the visualization
Unified trainable methodologies have led to
of SMPs. Combined with the evaluation of
many robust advances, including SCSI disks
agents, such a hypothesis constructs a novel
and information retrieval systems. This is
methodology for the simulation of forwarda direct result of the understanding of senerror correction.
sor networks. Given the current status of
autonomous information, system administraFuturists generally deploy the development
tors dubiously desire the emulation of the In- of write-ahead logging in the place of erasure
ternet, which embodies the unfortunate prin- coding. This is an important point to underciples of algorithms. Unfortunately, simu- stand. while conventional wisdom states that
lated annealing alone can fulfill the need for this challenge is regularly surmounted by the
extensible epistemologies.
synthesis of sensor networks, we believe that
We question the need for autonomous sym- a dierent solution is necessary. Thus, we see
1

Figure 6: Generated text : SCIgen Computer Science.

17

Comparison between inter-textual distance and other similarity index.

Figures 7,8 and 9 show the dendrograms obtained using cosine, Jaccard and Euclidean metrics.
They are computed using the R text mining package [18]. These dendrograms are to be compared to the one in figure 4. Dendrograms for Cosine and Euclidean do not group together the
Ike Antkare corpus.
Results, for the classification by assigning a text of the MLT corpus to the class of its nearest
neighbor, are given in table 4. The arXiv data set was not tested because of its size which make
the use of the R text mining package problematic.
Table 4: Classification of the MLT Corpus (122 papers) using Inter-textual distance, Cosine,
Euclidean and Jaccard metrics.
Non-SCIgen papers
wrongly classified

SCIgen papers
wrongly classified

Number of papers
well classified

Jaccard

121

Euclidean

30

92

Cosine

121

Inter-textual
Distance

122

0.3
l
l
l
I

l
l

0.2
II
lI
lll I

l
l ll

I
l
Il II

I
l

I Il l I
II
I
II I I I
I II I I
I
II I I
I
II
IIII
IIII I
I
I lII
l ll l
ll
ll l l l
I I l
lI I
II I
I I
II
II
I I I I
I
I
I
Il
I
Il
I I I
I
I I
lI I
I
I I
lI

ll

0.1

l l
l
|l
l||l
|l|
l||||
|l l|
| ||
l l||
|| |l
l| |
|||||l
l
l |l|
l l|l| |
|l l l
|
||| |
l | l|
l| |
l|l ||
ll|l
||
|l l|
ll
l l|l l l
|l |l
l l ll
l
l

0.0
l

l
|
||
||
||
||
||
||
||

Antkare

MLT

Figure 7: Cosine: dendrogram for analysis of corpora Antkare (black), Z (blue), MLT (red).

18

700

600

500

400

300

200

100

0.8

ll
l
l l
l
l
l l l

l
l

| |l|| |
|||
|||||
|
||| |
| | ||
||
|
|||
| || ||
||

|| |
|
| |l l|
|l
lll ll
l ll l
l l
l
l ll l
llll l
ll
ll l l
ll l
l l
|l l
|
l
|l l
l l ll
l l ll
l

0.6

||
| ||||
|
|| |
| |||
| ||

0.4

0.2

0.0

19

MLT

I I
II II I I
I II I
II I
I
III I I
I
I I II
II I
IIII I
I
II I I
II I
II
I II
III
I I IIII
I II I
I II
I
II II II
II
I
III I
I
II
I I
II
I ll l
l ll l

Antkare

Figure 9: Jaccard: Dendrogram for analysis of corpora Antkare (black), Z (blue), MLT (red).

lll

ll
ll
l|
ll
|l
ll|
||
|l
ll|
ll|
||
||l
|
l||
l||
ll|
||l
||
||
l|
ll
l||
ll|
||
||
|l|
|ll
||l
ll
ll
|l|
ll
ll
lll|
l
l
ll
l
lll ll
lll l
l
ll
l l ll
||
||
||
|
|||
||
||

Figure 8: Euclidean: dendrogram for analysis of corpora Antkare (black), Z (blue), MLT (red).

II I I
IIII I
I
I
II I
III
II I I
I I I
I
I
I
IIIII I
Il
l I
I lI
Il
I
I
I
I
II I I
I I II
I I
II I I
I II
I I II
I
I I II
I
III I
I
II I
I
l
l l

MLT

l l Il I
I
I
I
I

l
I

Antkare

Il