Monday, 16 February 2015

Academic Search Engine Spam and Google Scholar's Resilience Against it

 Source: http://quod.lib.umich.edu/j/jep/3336451.0013.305/--academic-search-engine-spam-and-google-scholars-resilience?rgn=main;view=fulltext



This paper was refereed by the Journal of Electronic Publishing’s peer reviewers.

Abstract

In a previous paper we provided guidelines for scholars on optimizing
research articles for academic search engines such as Google Scholar.
Feedback in the academic community to these guidelines was diverse. Some
were concerned researchers could use our guidelines to manipulate
rankings of scientific articles and promote what we call ‘academic
search engine spam’. To find out whether these concerns are justified,
we conducted several tests on Google Scholar. The results show that
academic search engine spam is indeed—and with little effort—possible:
We increased rankings of academic articles on Google Scholar by
manipulating their citation counts; Google Scholar indexed invisible
text we added to some articles, making papers appear for keyword
searches the articles were not relevant for; Google Scholar indexed some
nonsensical articles we randomly created with the paper generator
SciGen; and Google Scholar linked to manipulated versions of research
papers that contained a Viagra advertisement. At the end of this paper,
we discuss whether academic search engine spam could become a serious
threat to Web-based academic search engines.
Keywords: academic search engine spam, search engines, academic search engines, citation spam, spamdexing, Google Scholar

1 Introduction

Web-based academic search engines such as CiteSeer(X), Google Scholar, Microsoft Academic Search and SciPlore have introduced a new era of search for academic articles. In contrast to classic digital libraries such as IEEE Xplore, ACM Digital Library, or PubMed, Web-based academic search engines index PDF files of academic articles from any publisher that may be found on the Web.


Indexing academic PDFs from the Web not only allows easy and free
access to academic articles and publisher-independent search, it also
changes the way academics can make their articles available to the
academic community.


With classic digital libraries, researchers have no influence on
getting their articles indexed. They either have published in a
publication indexed by a digital library, and then their article is
available in that digital library, or they have not, and then the
article is not available in that digital library. In contrast,
researchers can influence whether their articles are indexed by
Web-based academic search engines: they simply have to put their
articles on a website to get them indexed.


Researchers should have an interest in having their articles indexed
by as many academic search engines and digital libraries as possible,
because this increases the articles’ visibility in the academic
community. In addition, authors should not only be concerned about the
fact that their articles are indexed, but also where
they are ranked in the result list. As with all search results, those
that are listed first, the top-ranked articles, are more likely to be
read and cited.


Furthermore, citation counts obtained from Google Scholar are
sometimes used to evaluate the impact of articles and their authors.
Accordingly, scientists want all articles that cite their articles to be
included in Google Scholar and they want to ensure that citations are
identified correctly. In addition, researchers and institutions using
citation data from Google Scholar should know how robust and complete
the data is that they use for their analyses.


In recent studies we researched the ranking algorithm of Google Scholar (Beel and Gipp 2009c, Beel and Gipp 2009a, Beel and Gipp 2009b) and gave advice to researchers on how to optimize their scholarly literature for Google Scholar (Beel et al. 2010). We called this method ‘Academic Search Engine Optimization’ (ASEO) and defined it as


“[...] the creation, publication, and modification of scholarly
literature in a way that makes it easier for academic search engines to
both crawl it and index it.” (Beel et al. 2010)


The idea of academic search engine optimization is controversial in
the academic community. Some researchers agree that scholars should be
concerned about it, and respond positively in various blogs and
discussion groups:


“In my opinion, being interested in how (academic) search engines
function and how scientific papers are indexed and, of course,
responding to these... well... circumstances of the scientific citing
business is just natural.” (Groß 2010)


“ASEO sounds good to me. I think it’s a good idea.” (Ian 2010)


“Search engine optimization (SEO) has a golden age in this internet
era, but to use it in academic research, it sounds quite strange for me.
After reading this publication [...] my opinion changed.” (Meskó 2010)


“This definitely needs publishing.” (Reviewer 2010)


Others argue against ASEO. Some of the critical feedback included statements like:


“I’m not a big fan of this area of research [...]. I know it’s in the call for papers, but I think that’s a mistake.” (Reviewer4 2009)


“[This] paper seems to encourage scientific paper authors to learn
Google scholar’s ranking method and write papers accordingly to boost
ranking [which is not] acceptable to scientific communities which are
supposed to advocate true technical quality/impact instead of ranking.” (Reviewer2 2009)


“[...] on first impressions [Academic Search Engine Optimization] sounds like the stupidest idea I’ve ever heard.” (Gunn 2010)


In our last paper (Beel et al. 2010) we concluded:


“Academic Search Engine Optimization (ASEO) should not be seen as a
guide how to cheat with search engines. It is about helping academic
search engines to understand the content of research papers, and thus
how to make this content more available.”


However, the concern that scientists might be tempted to
‘over-optimize’ their articles is at least worthy of investigation.
Therefore, we researched whether academic search engine spam can be
performed, how it might be done, and how effective it is. For us,
academic search engine spam (ASES) is the creation, modification, or
publication of academic articles as PDF files and resources related to
the articles, specially constructed to increase the articles’ or
authors’ reputations or ranking in academic search engines. Or, in
short, the abuse of academic search engine optimization techniques.


Initial results were published in a poster (Beel and Gipp 2010). The final results of our research are presented in this paper.


2 Research objective

The main objective of this study was to analyze the resilience of
Google Scholar against spam and to find out whether the following is
possible:


  • Performing citation spam to increase rankings, reputation, and visibility of authors and their articles.
  • Performing content spam to make papers appear in more search
    results, increasing their rankings and increasing authors’ publication
    lists.
  • Placing advertisement in PDFs.
In addition, we present our first ideas on how to detect and prevent
academic search engine spam. The results will help to answer the
following questions in further studies:


  • How reliable are Google Scholar’s citation counts, and should they be used to evaluate researcher and article impact?
  • To what extent can the ranking of Google Scholar be trusted?
  • To what extent can the linked content on Google Scholar be trusted?

3 Related work

To our knowledge, no studies are available on the existence of spam
in academic search engines or on how academic search engine spam could
be recognized and prevented. However, indexing and ranking methods of
Web-based academic search engines such as Google Scholar are similar to
those of classic Web search engines such as Google Web Search.
Therefore, a look at related work in the field of classic Web spam may
help in understanding academic search engine spam.


Most Web search engines rank Web pages based on two factors, namely
the Web page content and the amount (and quality) of links that point to
the Web page. Accordingly, Web spammers try to manipulate one or both
of these factors to improve the ranking of their websites for a specific
set of keywords. This practice is commonly known as ‘link spam’ and
‘content spam’.


Link spammers have various options for creating fraudulent links.
They can create dummy Web sites that link to the website they want to
push (link farms), exchange links with other webmasters, buy links on
third party Web pages, and post links to their websites in blogs or
other resources. Many researchers detected link spam (Gyöngyi and Garcia-Molina 2005, Benczur et al. 2005, Drost and Scheffer 2005, Fetterly et al. 2004, Benczúr et al. 2006, Saito et al. 2007, Wu and Chellapilla 2007, Gan and Suel 2007).


Content spammers try to make their websites appear more relevant for
certain keyword searches than they actually are. This can be
accomplished by taking content of other websites and combining different
(stolen) texts as ‘new content’, or by stuffing many keywords in a Web
page’s title, meta tags[1],
ALT-tags of images, and body text, or creating doorway pages, and
placing invisible text on a Web page. ‘Invisible text’ usually means
text in the same color as the background or in layers behind the visible
text. Again, much research has been performed to identify content spam (Urvoy et al. 2006, Nathenson 1998, Geng et al. 2008, Castillo et al. 2007).


A third type of Web spam is duplicate spam. Here, spammers try to get
duplicates of their websites indexed (and highly ranked). Figure 1
shows an example in which the three first results for a search query
point eventually to the same document. The chance that a Web surfer
would read the document is higher than if only one of the top results
had pointed to this paper[2]. Google provides guidelines for webmasters on how to avoid unintentional dupicate content spam[3]. Similar guidelines do not exist for Google Scholar.

Figure 1:: Example of duplicates on Google's result list (search query: 'tagging academic papers')
Figure 1:
Example of duplicates on Google's result list (search query: 'tagging academic papers')
Although Web spammers are continuously adjusting their methods and
developing new techniques (e.g. scraper sites, page hijacking, social
media spam, Wikipedia spam, and gadget spam), overall, search engines
are capable of fighting Web spam quite well.


Since academic search engines rank scientific articles in a similar
way as Web search engines rank Web pages, academic spam can be divided
into the same categories as Web spam: content spam, duplicate spam, and
link spam; however, in the case of academic papers ‘link spam’ is equal
to ‘citation spam.’


4 Motivation

Researchers could be tempted to do academic search engine spam for
several reasons: reputation, visibility, and ill will. We discuss these
reasons below.


4.1 Reputation

One reason researchers might perform academic search engine spam may
be to increase citation counts of their articles and hence enhance their
reputations. Citation counts are commonly used to evaluate the impact
and performance of researchers and their articles. In the past, citation
counts were amassed by organizations such as ISI’s Web of Science. Direct manipulation of Web of Science would be difficult, as ISI
checks citations in 10,000 journals from the reference lists in those
journals from 1900 to the present (and throws out duplicate references
in a single article). Nevertheless, some researchers are said to
manipulate their citation counts with citation circles, inappropriate
self-citations, etc.


Nowadays, citation counts from Web-based academic search engines are also used for impact evaluations. Software like Publish or Perish[4] and Scholarometer[5] calculate performance metrics such as impact factor and h-index (Hirsch 2005),
based on Google Scholar’s citation counts, to assist in analyzing the
impact of researchers and articles. These impact measures may be used to
support hiring and grants decisions.


We do not know to what extent these tools are used to evaluate the
performance of scientists. But several universities recommend Publish or Perish as an alternative to Web of Science (Harzing 2010) and many scholarly papers use citation data from Google Scholar for their analysis (Yang and Meho 2006, Harzing and van der Wal 2008, Kloda 2007, Bakkalbasi et al. 2006, Noruzi 2005, Meho and Yang 2007, Bar-Ilan 2007, Harzing and van der Wal 2008, Kousha and Thelwall 2008, Jacso 2008, Meho and Yang 2006, Moussa and Touzani 2009). Some evaluations even take into consideration download counts or the number of readers (Patterson 2009, Taraborelli 2010).


We believe that this kind of data will play an important role in
impact evaluations in the future. And the more these tools are used, the
higher the temptation for researchers to manipulate citation counts.


To increase their reputations and publication lists, researchers
might also try to create fake papers and get Google Scholar to index
these papers. A ‘fake paper’ could be any document that was solely
created for the purpose of manipulating citation counts, etc.


Researchers could try to modify articles by authors who are known in a
field, so that the articles reference the researchers’ articles or
appear to be co-authored by the nefarious researchers. Then it would
look as if an authority cited the manipulating researcher’s article or
as if the authority co-authored with the manipulating author.


Researchers are not the only ones who are evaluated by citation
counts; organizations such as universities or journals are evaluated the
same way and might therefore consider performing academic search engine
spam to increase their citation counts. One publisher has already been
caught putting pressure on authors to cite more articles from its
publications to increase the impact factor of the publishers’ journals (Havemann 2009).


4.2 Visibility

Researchers could duplicate one of their own articles with enough
slight changes and publish it on the Web to make the article appear new
to Google Scholar. If Google Scholar indexed it, the duplicate would
appear on Google Scholar as separate search result. Users would be more
likely read one of these articles than if only one result pointed to the
researcher’s work. The downside of this approach would be that real
citations would be divided among the various duplicates of the article.


Most academic search engines offer features such as showing articles
cited by an article, or showing related articles to a given article.
Citation spam could bring more articles from manipulating researchers
onto more of these lists. To do so, an author could modify an already
published article by inserting many additional references to papers
related to the modified paper. Authors of the cited papers would pay
attention to the modified article when they examine who is citing them,
and readers of the cited articles would more likely pay attention to the
citing article when they are searching for related work.


4.3 Ill-Will

Researchers might try academic search engine spamming just for fun,
or to damage others authors’ reputations by 'pushing' their article
rankings so obviously that the other authors are identified as spammers
by academic search engines and their articles are removed from the
index. On first glance, this idea might seem absurd. However, a similar
practice, called ‘competitor click fraud’, is common in paid search
results. Here, companies generate clicks on a competitor’s advertisement
to exhaust their budget (Wilbur and Zhu 2009, Soubusta 2008, Podobnik et al. 2006, Hadjinian et al. (2006), Gandhi et al. 2006).


A similar technique, deoptimization, is applied by so-called
‘webcare’ teams. These teams try to keep negative remarks and negative
publicity about a company from showing up high on search-engine results.
As a consequence, only positive websites appear high in the result
list.


4.4 Classic spam in academic articles

Classic non-academic spammers could place advertisement in
manipulated academic articles to generate revenue or create malicious
PDF files to either attack readers’ computers or attack the search
engines’ servers themselves. Just recently, Google and other companies
were attacked by hackers with malicious PDF files (Müll 2010).


5 Methodology

There are three basic approaches to academic search engine spam.


  • When creating an article, an author might place invisible text in
    it. This way, the article later might appear more relevant for certain
    keyword searches than it actually is.
  • A researcher could modify his own or someone else’s article and
    upload it to the Web. Modifications could include the addition of
    additional references, keywords, or advertisements.
  • A manipulating researcher could create complete fake papers that
    cite his or her own articles, to increase rankings, reputation, and
    visibility.
Over the past year, we performed several experiments on Google
Scholar. We placed invisible text in an article we published, modified
existing articles, and created several fake articles to test the
resilience of Google Scholar. The articles were uploaded to various
websites so Google Scholar could index them. Articles were uploaded to
our private homepage, http://beel.org; our project website, http://sciplore.org; the university website, http://www.ovgu.de; and to the social network websites http://mendeley.com, http://academia.edu and http://researchgate.net.


This paper should not be seen as a thorough experiment on how exactly
Google Scholar may be spammed. It is rather a case study and
proof-of-concept in which we perform various tests of how to spam Google
Scholar.


6 Results

6.1 Websites Google Scholar crawled

Google Scholar did not index our PDF files from mendeley.com and
researchgate.com, although other PDFs from those websites are indexed by
Google Scholar. PDFs from sciplore.org, beel.org and academia.edu were
indexed as well as PDFs from the university’s Web space.


6.2 Spamming while writing a real article

While writing one of our real papers (Beel and Gipp 2009b),
and before it was published, we added words in white color to the first
page (see Figure 2). In addition, we added several words in a layer
behind the original text (see Figure 3). Finally, a vector graphic, a
type of picture that can be searched and is machine readable, was
inserted. This vector graphic was also placed behind the original text,
and contained white text in a tiny font size (see Figure 4).


The paper then was submitted and accepted for a conference, published by IEEE, and included in IEEE Xplore. We did not let IEEE
know what we were doing, and the invisible text was not discovered.
About two months after publication the paper was crawled and indexed by
Google Scholar, which included the invisible text. That means users of
Google Scholar may find our article when they search for keywords that
appear only in the invisible text.

Figure 2:: White text on white background (highlighted for illustration)
Figure 2:
White text on white background (highlighted for illustration)
Figure 3:: Text in a hidden layer behind the original text (highlighted for illustration)
Figure 3:
Text in a hidden layer behind the original text (highlighted for illustration)
Figure 4:: The tiny white text right of the 'Vector graphic xxx:' is a vector grpahic (highlighed for illustration)
Figure 4:
The tiny white text right of the 'Vector graphic xxx:' is a vector grpahic (highlighed for illustration)

6.3 Modifying an already published article

6.3.1 Content modifications

We modified some articles we had already published and added
additional keywords (both visible and invisible) throughout the
document. Google indexed all modified PDFs and grouped them with the
original ones. That means users of Google Scholar may find these
modified articles when they search for the additional keywords. In other
words, researchers can make their articles appear for keyword searches
the original article would not be considered relevant for.


New keywords were also added to the PDF metadata (title and keyword
field). However, Google Scholar did not index the additional metadata.


6.3.2 Bibliography modifications

In several existing articles we added new references to the
bibliography. Some pointed to articles that were more recent than the
original article. These modified articles were uploaded to the Web, and
Google Scholar indexed all additional references. As a consequence,
citation counts and rankings of the cited articles increased.


That means researchers could easily increase citation counts and
rankings of their articles by modifying existing article (and not
necessarily their own). This way a researcher could also increase
visibility of his articles. He could modify one of his own articles, add
references to the bibliography, and the newly cited authors would then
probably pay attention to the article.


6.3.3 Adding advertisements

We modified one article (Beel and Gipp 2009b) and placed Viagra
advertisement in it, including a clickable link to the corresponding
website (see Figure 5). After a few weeks Google Scholar indexed the PDF
file and grouped it with the already indexed files.


That means users of Google Scholar interested in the full text of our research article (Beel and Gipp 2009b),
might download the manipulated PDF containing the Viagra advertisement
and we—if we were real spammers—could generate revenue from the
researchers visiting the advertised website.

Figure 5:: Viagra advertisement placed on the first page of an article with a link to a website selling Viagra
Figure 5:
Viagra advertisement placed on the first page of an article with a link to a website selling Viagra

6.4 Publishing completely new papers

So far, we had modified only existing papers. Google Scholar already
knew the articles’ metadata—title and author, for instance—when it was
indexing the manipulated PDFs.


We also made Google Scholar index papers that were never officially published.


6.4.1 Publishing nonsensical papers

Using the random paper generator SciGen (Stribling et al. 2005),
we created six random research papers. These papers consisted of
completely nonsensical text and bibliography. Only one real reference (Alcala et al. 2004)
was added. We created a homepage for a non-existent researcher and
offered the six created papers on this homepage for download. The
homepage was uploaded to the Web space OvGU.de, and linked by one of our
own homepages, so the Google Scholar crawler could find it.


Although Google Web Search indexed the homepage and PDFs after three
weeks, Google Scholar did not initially index the PDF files.

Figure 6:: The randomly created article 'A Case for Multicast Heuristic' with nonsensical text and uploaded to Academia.edu is indexed by Google Scholar and increased the citation count and ranking of our 'real' article.
Figure 6:
The randomly created article 'A Case for Multicast
Heuristic' with nonsensical text and uploaded to Academia.edu is
indexed by Google Scholar and increased the citation count and ranking
of our 'real' article.

We then uploaded one of the papers to Academia.edu. After two months Google Scholar indexed the paper from Academia.edu (see Figure 6) and from the university website as well, and ranking of the cited articles increased.


Apparently, Google Scholar has different trust levels for different
websites. It indexes unknown articles from the trusted websites, but
indexes only known articles from untrusted websites. In this case,
academia.edu seems to be considered trustworthy. Each article on that
platform is indexed by Google Scholar. It appears that once an article
is indexed from Academia.edu, other PDFs of that article are indexed, even from websites Google Scholar does not consider trustworthy.


6.4.2 Nonsensical text as real book

Recently created print-on-demand publishers such as Lulu, Createspace, and Grin
can publish a book, including ISBN, free, within minutes. We analyzed
whether a group of fake articles published as a real book would be
indexed by Google Scholar.


We created fourteen new fake articles with SciGen (Stribling et al. 2005).
We replaced the nonsense bibliography of each article with real
references. We bundled the fourteen articles in a single document and
published this document as a book with the publisher Grin (Beel 2009).
After a few weeks, the book was indexed by Google Books, and some weeks
later by Google Scholar. All fourteen articles can be found on Google
Scholar and their citations are displayed on Google Scholar too. That
means citation counts and rankings of around a hundred articles
increased because the fourteen fake papers cited these articles. Also
the (non-existent) authors are now listed in Google Scholar.


6.4.3 Publishing new articles based on real articles (duplicate spam)

In 2009 we published an article about how data retrieved from mind maps could enhance search applications (Beel et al. 2009).
It was titled ‘Information retrieval on mind maps—what could it be good
for?’ We took this article, changed the title to ‘Mind Maps and
Information Retrieval’ and replaced some references. The body text was
not changed. After uploading the article to the Web, Google Scholar
indexed it as a completely new article.


That means when users of Google Scholar search for ‘mind maps’ and
‘information retrieval’ the result set displays not only the original
article, but the modified one as well (see Figure 11). Accordingly, the
probability that users will read the article increases.

Figure 7:: Duplicates with identical content but different title are listed as separate search results
Figure 7:
Duplicates with identical content but different title are listed as separate search results
Something similar happened with a book we published about rewarding project teams (Beel 2007).
Google Scholar indexed the original print version, which is also
available on Google Books. When we posted the PDF on the book’s website,
http://team-rewards.de,
Google Scholar indexed it as a new article. Differences between the
documents, each about 100 pages, are minimal. However, as Figure 8
shows, Google Scholar has misidentified the title. The correct title is
on Google Books: ‘Project Team Rewards: Rewarding and Motivating your
Project Team’. The PDF’s title was incorrectly identified as ‘Project
Team Rewards’.

Figure 8:: Multiple indexing of the same document
Figure 8:
Multiple indexing of the same document
As a consequence of this misidentification, both documents are
displayed for searches for the term ‘project team rewards’ or other
similar terms. In addition, the cited articles all received two
citations because the original book and the PDF from the website were
indexed separately.


Based on these results, it seems that Google Scholar is using only a
document’s title to distinguish documents. If titles differ, documents
are considered different.


6.5 Miscellaneous

In our research we saw some issues that might be relevant in
evaluating Google Scholar’s ability to handle spam and its reliability
for citations counts.


6.5.1 Value of citations

Google Scholar indexes documents other than peer-reviewed articles. For instance, Google Scholar has indexed 4,530 PowerPoint presentations[6] and 397,000 Microsoft Word
documents. It has indexed a Master thesis proposal from one of our
students and probably many proposals more. Citations in all these
documents are counted[7]. It is apparent that a citation from a PowerPoint
presentation or thesis proposal has less value than a citation in a
peer reviewed academic article. However, Google does not distinguish on
its website between these different origins of citations[8].


6.5.2 Wikipedia articles on third party websites

Google Scholar indexes Wikipedia articles when the article is available as PDF on a third party website. For instance, the Wikipedia article on climate change[9] is also available as a PDF on the website http://unicontrol-inc.com (with a different title). Google Scholar has indexed this PDF (see Figure 9) and counted its references.

Figure 9:: Indexed Wikipedia article from third party website
Figure 9:
Indexed Wikipedia article from third party website
That means, again, that not all citations on Google Scholar are what
we call ‘full-value’ citations. More importantly, researchers could
easily perform academic search engine spam just by citing their papers
in Wikipedia articles, creating a PDF of the Wikipedia article, and uploading the PDF to the Web.


6.5.3 PDF duplicates / PDF hijacking

Google Scholar indexes identical PDF files that have different URLs
separately, even if they are on the same server. In case of our article
‘Google Scholar’s Ranking Algorithm: An Introductory Overview’, four
PDFs on the domain beel.org (see Figure 10) were all indexed. Google
even considers the same PDF with same URL—once with and once without www—as different.


That means a spammer could upload the same PDF several times to the
same Web page and all PDFs would be displayed on Google Scholar.
Consequently, the probability that a user downloads the manipulated PDF
would increase.

Figure 10:: Indentical PDFs from teh domain beel.org grouped as separate versions.
Figure 10:
Indentical PDFs from teh domain beel.org grouped as separate versions.
The ranking of grouped PDFs depends mainly on the file date—newer
files are listed higher. That means spammers publishing modified
versions of an article most likely will see their manipulated PDF as the
primary download link for an article. This was also the case in our
test with the manipulated PDF containing Viagra advertisement. The
manipulated PDF is the most current PDF and displayed as primary
download link (see Figure 11).

Figure 11:: Ranking of muliple PDF files
Figure 11:
Ranking of muliple PDF files
A similar practice is known from Web spam. ‘Page hijacking’ describes
the practice that spammers create Web pages (with advertisements,
malicious code, etc.) similar to a popular website. Under some
circumstances Google identifies the duplicate as the original Web page
and displays the duplicates’ website as the primary search result.


6.5.4 Misidentification of journal name

By coincidence we realized that it is possible to manipulate the
journal name Google Scholar anticipates as the publishing journal of an
article. One of our papers (Gipp and Beel 2009) includes a vector graphic on the second page that illustrates how recommendations are made on our website http://sciplore.org.
This vector graphic includes bibliographic information, among others
‘Epidemiology, vol. 19, no. 3’ (see Figure 12 for a screenshot of that
PDF and the vector graphic).

Figure 12:: PDF with a vector graphic showing a popular journal name (Epidemiology)
Figure 12:
PDF with a vector graphic showing a popular journal name (Epidemiology)
Interestingly, Google Scholar used this bibliographic information as
the name of the journal our article was published in (although it was
not). A search on Google Scholar for our article shows the article as
being published in Epidemiology, a reputable journal by the publisher JSTOR (see Figure 13).

Figure 13:: Misidentification of journal name
Figure 13:
Misidentification of journal name
Apparently, Google Scholar is using text within an article to
identify the article’s publishing journal. This could be used by
spammers to make their papers appear as if they were being published in
reputable journals.


7 Discussion

We discussed academic search engine spam with several colleagues.
Some congratulated us on our work; others considered it to be
meaningless or even negative for the academic community. Apparently,
opinions vary strongly about academic search engine spam. Therefore, we
believe that academic search engine optimization and the potential
threat of abusing it should be discussed.


We have heard the argument that academic spam might be a less serious
threat to academic search engines than Web spam is to Web search
engines. First, the effort required for academic search engine spam is
high in contrast to the effort required for normal Web spam. Creating
spam Web pages, including the registration of new domains, can be done
almost automatically within seconds. In contrast, creating modified PDFs
or publishing articles with print-on-demand publishers requires
significantly more time.


Second, the benefit of spam for researchers is not as immediate and
measurable as it is for other Web spammers. While a Web spammer can
expect a certain amount of money for each additional visitor, a
researcher can hardly specify the benefit of additional citations and
readers.


Finally, and most importantly, researchers are not anonymous. In Web
search, a website’s domain might be banned by the search engine if the
site is identified as spam but the spammer could register a new domain
within seconds (with a fake identity, if necessary). In contrast,
researchers need to think about their reputation. If a researcher doing
academic search engine spam were exposed, the academic search engine
would ban all his articles permanently, and his reputation in the
academic community would likely be permanently damaged.


However, although the vast majority of researchers are honest, it is
widely known that there are some researchers performing unethical and
even illegal actions to increase their reputation (see, e.g., [Judson 2004]
for examples). Therefore, it must be assumed that some researchers are
willed to do academic search engine spam, despite the risks.


Also journals and conferences might be tempted to do academic search
engine spam. Most, if not all, journal and conference rankings consider
citation counts as the major or even only factor for calculating the
ranking. By citation spam, journals and conferences could dramatically
increase their rankings, and therefore, most likely, their revenue.


Journals might also be tempted to perform academic search engine spam to attract more visitors to their websites. The publisher SAGE states that 60% of all online readers come via Google and Google Scholar to their journals (SAGE 2010).
This percentage may increase in a few years. Therefore, academic search
engine spam could bring thousands of new visitors and potential
revenue. Small and currently unknown journals and conferences might be
especially willing to take the risk. If they are discovered, they could
found another journal or conference and try again.


Maybe most importantly, ‘normal’ Web spammers probably will place
their spam in modified research articles as soon as they learn that it
is possible. Google Scholar provides a new platform to them with hardly
any barriers to distributing their spam. There is no reason to assume
that normal spammers would not take advantage of this.


Most publishers seem not to be aware of
the possibility of academic search engine optimization and academic
search engine spam. We scrutinized publishing policies of three major
publishers in the field of computer science (IEEE, Springer, and ACM) and could not find any rules or policies that address things like including invisible text.


Some publishers are aware of the benefits of academic search engine optimization. The publisher SAGE, for instance, suggests the following practice for authors:


“Search engines look at the abstract page of your article [...] Try
to repeat the key descriptive phrases [but] don’t overplay it, focus on
just 3 or 4 key phrases in your abstract.” (SAGE 2010)


“Ensure the main key phrase for your topic is in your article title.” (SAGE 2010)


This advice does not differ significantly from the guidelines we provided in (Beel et al. 2010). However, some journals’ recommendations cross what we would consider legitimate ASEO. For instance, the Journal of Information Assurance and Security (JIAS) gave the following ‘recommendation’ to us after a paper we submitted in 2009 was accepted:


“Please [...] improve the introduction/related research section by including all the past related papers published in JIAS.”


Also the International Journal of Web Information Systems
recommended that we “add references from papers previously published in
International Journal of Web Information Systems” after one of our
papers was accepted in 2010.


To us, the intention of these recommendations seem primarily to be to
increase citation counts of the journal and hence to improve metrics
such as the impact factor[10].


8 Conclusion

As long as Google Scholar applies only very rudimentary or no
mechanisms to detect and prevent spam, citation counts should be used
with care to evaluate articles’ and researchers’ impact. Similarly,
researchers should be aware that rankings and linked content might be
manipulated. Overall, Google Scholar is a great tool that may help
researchers find relevant articles. However, Google Scholar is a
Web-based academic search engine and as with all Web-based search
engines, the linked content should not be trusted blindly.


To academic search engines we suggest applying at least the most
common spam detecting techniques known from Web search engines. They
include analyzing documents for invisible text and either ignoring this
text or ignoring the entire document. Also, very small fonts, especially
in vector graphics, should not be indexed. With common spam detection
methods the PDFs could also be analyzed for ‘normal’ spam. If an
authoritative article directly from the publisher is available, only
citations from this article should be counted, and not form other
versions of the article found on the Web. It is also questionable
whether counting citations from PowerPoint slides and Microsoft Word documents is sensible.


In addition, documents should be analyzed for ‘sense making’. The documents we created with SciGen and published with Grin and on Academia.edu
consisted of completely nonsensical text, but still they were indexed.
Articles with identical or nearly identical text but different titles
should not be listed as separate search results but should be grouped.
Also, identical PDFs, especially when they are from the same domain,
should not be listed as separate versions.


Finally, we suggest that publishers change their policies:
over-optimization of articles should be a violation of their policies
and lead to appropriate consequences. However, the academic community
needs to decide what actions are appropriate and when academic search
engine optimization ends and academic search engine spam begins.


9 Summary

Our study on the resilience of Google Scholar delivers surprising
results: Google Scholar is far easier to spam than the classic Google
Search for Web pages. While Google Web Search is applying various
methods to detect spam and there is lots of research on detecting spam
in Web search, Google Scholar applies only very rudimentary
mechanisms—if any—to detect spam.


Google Scholar indexed invisible text in all kind of articles. A
researcher could put invisible keywords in his article before, or even
after, publication and increase the ranking and visibility of this
article on Google Scholar.


Google Scholar counted references that were added to modified
versions of already published articles. That means authors could add
references in their articles after official publication. If these
altered articles were published on the Web, Google would index them.
This way, researchers could increase citation counts and rankings of the
cited articles. They could also bring attention to their articles
because the cited authors might investigate who has cited them.
Researchers could also modify articles from other authors and add
references to their own articles. This way, scholars could create the
impression that an authority in their field cited their articles and
increase citation counts as well.


Google Scholar also indexed fake articles uploaded to trusted sources such as Academia.edu and articles that were published as book with a print-on-demand publisher such as Grin[11].
This gives researchers another way to manipulate citation counts and
extend their publication lists. An author could create a fake article
with his or her name and the name of a popular researcher as co-author.
This method could also be used to publish a real article again but with a
different title, so the different variations would appear as separate
items in the result lists (duplicate spam).


Google Scholar is indexing file formats other than PDFs, such as PowerPoint presentations (.ppt) and Microsoft Word
documents (.doc), and counting references that were made in these
files. Although we did not test it, one might assume that it would be
easy to create PowerPoint presentations and
doc files citing a specific article just with the intention of pushing
the article’s ranking. Google Scholar is also indexing non-peer-reviewed
academic documents such as thesis proposals or Wikipedia articles offered on third party websites.


It was also easy to perform duplicate spam. With changed titles,
basically identical PDFs were identified as separate articles. In
addition, Google Scholar seems to rank new PDFs higher than older PDFs.
That means manipulated PDFs most likely would appear as the primary
download link.


By coincidence we realized that Google Scholar assigned a paper to a
journal named in the full text of the article. We did not investigate
this further, but it might be possible to make an article seem to have
been published in a reputable journal although it never was.


Finally, Google Scholar indexed modified versions of articles that
contained advertisements. Certainly, researchers would not add
advertisement to their own articles. But it is imaginable that normal
spammers could download thousands of academic PDFs, automatically place
their advertisement in these PDFs, and upload them to the web. Google
Scholar would index them, and users of Google Scholar interested in an
article’s full text might download these modified articles and see the
advertisement.


Some might argue that academic search engine spam is a less serious
threat to academic search engines than classic Web spam is to Web search
engines. However, the potential benefits of academic search engine spam
might be too tempting for some researchers. In addition, we see little
reason why normal Web spammers should not place their advertisement in
academic articles.


To prevent academic search engine spam, Google Scholar (and other
Web-based academic search engines) should apply at least the common spam
detection techniques known from Web spam detection, analyze text for
sense-making, and not count all citations.


Note

We would like to note that the intention of this paper was not to
expose Google Scholar. The intention was to stimulate a discussion about
academic search engine optimization and the threat of academic search
engine spam. We chose Google Scholar as the subject of our study because
Google Scholar probably is the best and largest academic search engine
indexing PDFs from the Web. Currently, we are developing our own
academic search engine, SciPlore (http://sciplore.org). As yet, SciPlore has no protections against spam either. A very brief investigation of CiteSeer and Microsoft Academic Search indicates that they do not detect academic search engine spam either.


Acknowledgements

We thank Bert van Heerde from Insyde for his valuable feedback.



Notes

  1. Meta tags are rarely used by spammers since most search engines ignore meta tags due to spam issuesreturn to text
  2. We do not claim that the
    author of the example paper did duplicate spam. It is likely that Google
    was not able to identify the different pages as duplicates. However,
    this illustrates what duplicate spam might look like. return to text
  3. The amount of indexed files of a certain type (e.g. ppt) are identifiable via the search query “filetype:ppt” return to text
  4. We took a sample of 10
    presentations and Microsoft Word documents that contained citations and
    all citations in these files were counted.return to text
  5. It could be that Google
    Scholar weights citations differently when using them for ranking
    articles. However, third parties parsing Google Scholar cannot identify
    any distinctions.return to text
  6. Due to these recommendations we decided to withdraw the submitted papers.return to text
  7. It has to be mentioned
    that we published this book under our real names. However, it would have
    been just as easy to publish it with a fake identity (though this would
    have violated the terms of service of the print on demand publisher).return to text

References

Alcala, F., J. Beel, A. Frenkel, B. Gipp, J. Lülf, and
H. Höpfner. 2004. "UbiLoc: A System for Locating Mobile Devices using
Mobile Devices." In K. Kyamakya, ed., Proceedings of 1st Workshop on Positioning, Navigation and Communication 2004 (WPNC 04): 43–48. University of Hanover. Also available on http://beel.org.
Bakkalbasi, N., K. Bauer, J. Glover, and L. Wang. 2006.
"Three options for citation tracking: Google Scholar, Scopus and Web of
Science." Biomedical Digital Libraries 3.
Bar-Ilan, J. 2007. "Which h-index? - A comparison of WoS, Scopus and Google Scholar." Scientometrics 74(2):257–271. doi:10.1007/s11192-008-0216-y
Beel, J. 2007.Project Team Rewards - Rewarding and Motivating your Project Team. Scotts Valley (USA): Createspace. ISBN 978-1434816269. Also available on http://project-team-rewards.com.
Beel, J., ed. 2009. Computer Science: New Generations. Munich (Germany): Grin Verlag. ISBN 978-3-640-32875-8.
Beel, J. and B. Gipp. 2009a. "Google Scholar’s Ranking Algorithm: An Introductory Overview." In B. Larsen and J. Leta, eds., Proceedings of the 12th International Conference on Scientometrics and Informetrics (ISSI’09) 1: 230–241. Rio de Janeiro: International Society for Scientometrics and Informetrics. ISSN 2175-1935. Also available on http://www.sciplore.org.
Beel, J. and B. Gipp. 2009b. "Google Scholar’s Ranking
Algorithm: The Impact of Articles’ Age (An Empirical Study)." In
S. Latifi, ed., Proceedings of the 6th International Conference on Information Technology: New Generations (ITNG’09): 160–164, Las Vegas: IEEE. ISBN 978-1424437702. Also available on http://www.sciplore.org.
Beel, J. and B. Gipp. 2009c. "Google Scholar’s Ranking
Algorithm: The Impact of Citation Counts (An Empirical Study)." In
A. Flory and M. Collard, eds., Proceedings of the 3rd IEEE International Conference on Research Challenges in Information Science (RCIS’09): 439–446. Fez (Morocco):IEEE. ISBN 978-1-4244-2865-6. Also available on http://www.sciplore.org.
Beel, J. and B. Gipp. 2010. "On the Robustness of Google Scholar Against Spam." In Proceedings of the 21th ACM Conference on Hyptertext and Hypermedia: 297–298. Toronto: ACM. Also available on http://www.sciplore.org.
Beel, J., B. Gipp, and J. O. Stiller. 2009. "Information Retrieval on Mind Maps – What could it be good for?" In Proceedings
of the 5th International Conference on Collaborative Computing:
Networking, Applications and Worksharing (CollaborateCom ’09)
: 1–4. Washington: IEEE. ISBN 978-963-9799-76-9. Also available on http://www.sciplore.org.
Beel, J., B. Gipp, and E. Wilde. 2010. "Academic Search
Engine Optimization (ASEO): Optimizing Scholarly Literature for Google
Scholar and Co." Journal of Scholarly Publishing41(2): 176–190. Also available on http://www.sciplore.org. doi:10.3138/jsp.41.2.176
Benczúr, A., K. Csalogány, and T. Sarlós. 2006. "Link-based similarity search to fight web spam." Adversarial Information Retrieval on the Web (AIRWEB), Seattle, Washington, USA.
Benczur, A., K. Csalogány, T. Sarlós, and M. Uher. 2005. "SpamRank – Fully Automatic Link Spam Detection." In Adversarial Information Retrieval on the Web (AIRWEB ’05).
Castillo, C., D. Donato, A. Gionis, V. Murdock, and
F. Silvestri. 2007. "Know your neighbors: Web spam detection using the
web topology." In Proceedings of the 30th annual international ACM SIGIR conference on Research and development in information retrieval, 430. ACM.
Drost, I. and T. Scheffer. 2005. "Thwarting the nigritude ultramarine: Learning to identify link spam." Lecture Notes in Computer Science 3720:96. doi:10.1007/11564096_14
Fetterly, D., M. Manasse, and M. Najork. 2004. "Spam, damn
spam, and statistics: Using statistical analysis to locate spam web
pages." 1–6.
Gan, Q. and T. Suel. 2007. "Improving web spam classifiers using link structure." In Proceedings of the 3rd international workshop on Adversarial information retrieval on the web: 20. ACM.
Gandhi, M., M. Jakobsson, and J. Ratkiewicz. 2006. "Badvertisements: Stealthy click-fraud with unwitting accessories." Journal of Digital Forensic Practice1(2):131–142.doi:10.1080/15567280601015598
Geng, G., C. Wang, and Q. Li. 2008. "Improving Spamdexing Detection Via a Two-Stage Classification Strategy." 356.
Gipp, B. and J. Beel. 2009. "Identifying Related Documents
For Research Paper Recommender By CPA And COA." In S. I. Ao,
C. Douglas, W. S. Grundfest, and J. Burgstone, eds., International Conference on Education and Information Technology (ICEIT ’09), Lecture Notes in Engineering and Computer Science
1:636–639. Berkeley (USA): International Association of Engineers
(IAENG): Newswood Limited. ISBN 978-988-17012-6-8. Also available on http://www.sciplore.org.
Groß, M. 2010. "academic search engine optimization." February 10. Blog.
Gunn, W. 2010. Feedback on ASEO. January. Blog Comment.
Gyöngyi, Z. and H. Garcia-Molina. 2005. "Link spam alliances." In Proceedings of the 31st international conference on Very large data bases: 528. VLDB Endowment.
Hadjinian, D. et al. 2006. "Clicking away the competition:
The legal ramifications of click fraud for companies that offer pay per
click advertising services." Shidler JL Com. & Tech. 3:5–16.
Harzing, A. and R. van der Wal. 2008. "Google Scholar as a new source for citation analysis." Ethics in Science and Environmental Politics (ESEP) 8(1):61–73. doi:10.3354/esep00076
Harzing, A. and R. van der Wal. 2008. "A Google Scholar
h-index for journals: An alternative metric to measure journal impact in
economics and business." Journal of the American Society for Information Science 60(1):41–46.doi:10.1002/asi.20953
Harzing, A. W. 2010. Publish or Perish in the news. Website.
Havemann, F. 2009. Einführung in die Bibliometrie. Humboldt University of Berlin.
Hirsch, J. E. 2005. "An index to quantify an individual’s scientific research output." Proceedings of the National Academy of Sciences 102(46):16569–16572. doi:10.1073/pnas.0507655102
Ian. 2010, January. Feedback on ASEO. Blog Comment.
Jacso, P. 2008. "Testing the calculation of a realistic
h-index in Google Scholar, Scopus, and Web of Science for FW Lancaster."
Library Trends 56(4):784–815. doi:10.1353/lib.0.0011
Judson, H. 2004. The great betrayal: fraud in science. Houghton Mifflin Harcourt (HMH).
Kloda, L. 2007. "Use Google Scholar, Scopus and Web of Science for comprehensive citation tracking." Evidence Based Library and Information Practice2(3):87.
Kousha, K. and M. Thelwall. 2008. "Sources of Google
Scholar citations outside the Science Citation Index: A comparison
between four science disciplines." Scientometrics74(2):273–294. doi:10.1007/s11192-008-0217-x
Meho, L. and K. Yang. 2006. "A new era in citation and bibliometric analyses: Web of Science, Scopus, and Google Scholar." Arxiv preprint cs/0612132.
Meho, L. and K. Yang. 2007. "Impact of data sources on
citation counts and rankings of LIS faculty: Web of Science vs. Scopus
and Google Scholar." Journal of the American Society for Information Science and Technology58:2105–25. doi:10.1002/asi.20677
Meskó, B. 2010. "Academic Search Engine Optimization in Google Scholar." January. Blog.
Müll, D. 2010. "Der Cyberkrieg hat begonnen: 34 US-Unternehmen teils stark von Attacken getroffen." January. News Website.
Moussa, S. and M. Touzani. 2009. "Ranking marketing journals using the Google Scholar-based hg-index." Journal of Informetrics.
Nathenson, I. 1998. "Internet infoglut and invisible ink: Spamdexing search engines with meta tags." Harv. J. Law & Tec 12:43–683.
Noruzi, A. 2005. "Google Scholar: The new generation of citation indexes." Libri 55(4):170–180. doi10.1515/LIBR.2005.170
Patterson, M. 2009. "Article Level Metrics." September. Blog.
Podobnik, V., K. Trzec, and G. Jezic. 2006. "An auction-based semantic service discovery model for e-commerce applications." Lecture Notes in Computer Science4277:97. doi:10.1007/11915034_32
Reviewer, J. 2010. Your Submission: Academic Search Engine Spam. March. Private Communication (Email).
Reviewer2, C. 2009. Feedback I on Reverse Engineering
Google Scholar’s Ranking Algorithm (Conference Reviewer). March. Private
Communication (Email).
Reviewer4, C. 2009. Feedback II on Reverse Engineering
Google Scholar’s Ranking Algorithm (Conference Reviewer). March. Private
Communication (Email).
SAGE. 2010. "Authors - How to help readers find your article online." Website.
Saito, H., M. Toyoda, M. Kitsuregawa, and K. Aihara. 2007. "A large-scale study of link spam detection by graph algorithms." In Proceedings of the 3rd international workshop on Adversarial information retrieval on the web: 48. ACM.
Soubusta, S. 2008. On Click Fraud. Information Wissenschaft und Praxis59.2:136.
Stribling, J., M. Krohn, and D. Aguayo. 2005. Scigen-an automatic cs paper generator.
Taraborelli, D. 2010. ReaderMeter: Crowdsourcing research impact. September. Blog.
Urvoy, T., T. Lavergne, and P. Filoche. 2006. "Tracking web spam with hidden style similarity." In AIRWeb 2006:25.
Wilbur, K. and Y. Zhu. 2009. "Click fraud." Marketing Science 28(2):293–308. doi:10.1287/mksc.1080.0397.
Wu, B. and K. Chellapilla. 2007. "Extracting link spam using biased random walks from spam seed sets." In Proceedings of the 3rd international workshop on Adversarial information retrieval on the web:44. ACM.
Yang, K. and L. Meho. 2006. "Citation analysis: A comparison of google scholar, scopus, and web of science":43.


Joeran Beel is a visiting scholar at UC Berkeley and PhD student in
computer science at OvGU Magdeburg (Germany). He obtained an MSc in Business Information Systems
at OvGU Magdeburg and graduated with distinction and was cited as the
computer science department’s best student. In addition, he obtained an
MSc in Project Management at Lancaster University Management School (UK).
In recent years he published several papers about academic search engines. He is a co-founder of the academic search engine SciPlore (www.sciplore.org) and the machine readable digital library Mr. dLib (www.mr-dlib.org).
He won several awards for his research, including one from Germany’s
Chancellor Gerhard Schröder. Contact details may be found on his
homepage (www.beel.org).



At the age of 15 Bela Gipp won several prizes at Jugend Forscht,
Germany’s premier national youth science competition. The German
Chancellor honored him after he took first prize in the state-level
round for the third time. Scholarships allowed him to study in
Australia, England, and China, and at UC Berkeley, in California. After
obtaining his master’s in the field of computer science and an MBA, he
started working for SAP AG in a research joint venture.
Currently he is a visiting researcher at UC Berkeley, where he works on
his doctoral research. Central topics are network theory and
bibliometric analysis. Besides his theoretical research, he develops
open-source software for scientists as a founder of SciPlore.org.
Publications can be found on his website www.gipp.com.




Academic Search Engine Spam and Google Scholar's Resilience Against it

No comments:

Post a Comment