SMART WEB VISIBILITY OF ORGANIZATIONS
Augusto Klinger, José Valdeni de Lima and José Palazzo Moreira de Oliveira
Instituto de Informática, UFRGS, Av. Bento Gonçalves 9500, Porto Alegre, Brazil
Keywords: Web Visibility, Smart Web Visibility, Universities Ranking, Homonymous Problem.
Abstract: Smart Web Visibility, in this paper, is the study of measure techniques used to retrieval information about-
words, expressions or terms (for example, acronyms) on the web. The visibility has straight relation with
presentation order of information retrieved. We consider the Smart Web Visibility as a subfield of the We-
bometrics, the same as Web Visibility is its subfield too. Our approach is based on different rankings from
different search engines to evaluate the Smart Web Visibility by processing the homonymous problem be-
fore scoring. We begin with original results of search engines output, to emphasise after with methods to
adding semantics to the queries. Finally, to demonstrate the viability of our ideas we employed acronyms of
Brazilian universities to evaluate the smart visibility and compare with the actual situation of Brazil univer-
sities published by Webometrics Ranking. The main contribution of this work is a new way to evaluate the
Web Visibility, named Smart Web Visibility, which shows how the universities are ranked by multiple
search engines.
1 INTRODUCTION
Visibility on the Web is a quantitative measure of
the visibility of a webpage on the network, deter-
mined by the ease that Web users find it. The study
area of measures of visibility on the Web is named
Web Visibility considering the quantitative aspects
of construction and use of information on the Web
viewed under bibliometric aspects (Björneborn and
Ingwersen, 2004). The goal of Webometrics is to ob-
tain information by measurements on the various as-
pects of the Web obtaining, for example, statistics
data about popularity, clusters of websites and dis-
tribution of information. We named Smart Web Vi-
sibility the visibility provided by the different vi-
sions of the eyes of the Web, the search engines.
Web visibility measuring is important in several
respects, principally to evaluate the level of advertis-
ing or measure the impact of a trademark, product or
institution in the network. There are different ways
of measuring the visibility of a website: number of
hits, number of links that lead to the website, or po-
sition in the ranking of a search engine.
The measure most spread in the literature is the
use of the number of web links pointing to a web-
page, or inlinks, as a measure of visibility in the
network (Aguillo, Granadino, Ortega and Prieto,
2006; Aguillo, Granadino and Ortega, 2006). This
measure is also used to generate the ranking of uni-
versities by the Cybermetrics Lab, the Webometrics
Ranking of World Universities. The criterion of
number of inlinks is usually used in the algorithms
that assemble the rankings of search engines, as the
world famous PageRank (Page, Brin, Motwani and
Winograd, 1999). Few studies use the number of
accesses to a webpage as a measure of visibility
(Aaltojärvi, Arminen, Auranen and Pasanen, 2008)
and there isn’t widespread works using search
engine’s results for the calculation of visibility on
the web, although they are the main option for any
web search. The development of an approach to
evaluate visibility based on web search engines is
presented as an interesting alternative to the existing
evaluation web visibility methods. The measure-
ments are relevant not only to the field of web
marketing, but also for the elaboration of rankings in
certain domains and evaluations of popularity of
general purpose on the web.
2 RELATED WORK
A work of Aguillo (2006b) analyzed the presence of
Brazilian universities on the web. According to the
author, the developing countries of Latin America
are making efforts to publish electronically the re-
sults of their researches and studies. The size of their
web domains has grown, as well as its visibility on
671
Klinger A., Valdeni de Lima J. and Palazzo Moreira de Oliveira J..
SMART WEB VISIBILITY OF ORGANIZATIONS.
DOI: 10.5220/0003928906710676
In Proceedings of the 8th International Conference on Web Information Systems and Technologies (WEBIST-2012), pages 671-676
ISBN: 978-989-8565-08-2
Copyright
c
 2012 SCITEPRESS (Science and Technology Publications, Lda.)

the web. The results show an increasingly strong
presence of Brazilian universities on the web, but
still far from developed countries. The study data
were obtained from eight search engines. The indi-
cators were the number of pages (size), number of
inlinks (visibility) and number of visits (popularity).
The São Paulo University (USP) has led the rank-
ings according to three indicators. University of
Campinas (UNICAMP) was second in size and visi-
bility, the second in popularity is the Federal Uni-
versity of Rio de Janeiro (UFRJ).
Web visibility indicators were examined to as-
sess how far the collaboration in science and tech-
nology publications are visible on the Web (Aguillo
and Kretschmer, 2004). The study found that about
80% of the bibliography with multiple authors is vis-
ible through search engines. The studies by Kret-
schmer (2007) show that the structures of hyperlinks
does not reflect the collaborative structures of bibli-
ographic data, but web visibility indicators are dif-
ferent from hyperlinks and can be used successfully
as indicators of web collaboration.
Barjak and Thelwall (2008) report the results of
a study of the connection between the count of in-
links and the real significance of the websites of
about 400 research groups in Europe. The analysis
confirmed that the size of research groups and their
presence on the Web are important to attract links,
while the scientific production itself is not. The in-
terpretation of data from search engines need to be
further studied before to take conclusions about its
usefulness as indicators, conclude the authors.
However data from search engines are widely
used in several works with different purposes. A
quick web search reveals many applications such as:
domain evaluators (Dnscoop, 2009; Cubestat, 2008),
which provide a dollar amount to the target site;
trees of words, which are based on user queries of
search engines (Viegas, n.d.).
The works of Espadas, Calero and Piatinni
(2008) and Gori and Witten (2005) highlight impor-
tant points of Web Visibility. The first deals with the
problem that search engines do not make large parts
of the Web visible, proposing a method for indexing
sites. The second explains some heuristics adopted
by search engines and expose their weakness by al-
lowing the construction of artificial communities of
sites that link to each other in order to improve their
rankings. A possible solution to the problem is in the
Semantic Web.
A previous work used notions of relevance and
precision of metasearch engine rankings combined
with a rankings fusion method to develop a calcula-
tion of Web Visibility (Klinger et. al., 2011). The re-
sults serve as an indication that the web search en-
gines provide interesting data for the evaluation of
visibility and point to future studies to apply and ex-
pand the formula in a general way.
3 METASEARCH ON THE WEB
This work aims to define a new way to evaluate vi-
sibility on the Web based on information collected
by several search engines. As the volume of data on
the Web is very large, and growing, no search en-
gine can index the entire web. Additionally, only the
best placed websites are displayed to users in effi-
ciency concern. What we have as the result of a
search for a particular term in any web search engine
is a classification according to the search engine
used, covering a portion of the web, classified ac-
cording to their heuristics, techniques and proprie-
tary algorithms. One way to increase the scope of
coverage of the Web is the utilization of more search
engines. This process of consulting several search
engines is known as metasearch. As the search en-
gines are the dominant points of access to webpages,
the metasearch engine presents itself as an interest-
ing tool for measuring visibility on the web, result-
ing in a quantitative data representing how accessi-
ble and how well regarded is the website in accor-
dance with the web search engines. Using more
search engines, tends to increase the diversity of cri-
teria and range, generating a more reliable value of
web visibility.
Considering the official website of an organiza-
tion as the major milestone of its presence on the
network, is expected that when a search engine is
consulted for the organization name, or its acronym,
the official website is between the first placed re-
sults. This means that the organization website, and
therefore its name, has a good visibility under the
search engine used. By applying a metasearch with
any term (or list of terms) the results are different
rankings, one for each search engine to which the
metasearch system forwarded the query. Usually a
single ranking is displayed to the user, making ne-
cessary a fusion technique for the various rankings.
There are several methods to merge different classi-
fication functions, Rankings Fusion. However, as we
haven’t interest in see a ranking of all the webpages
retrieved, we want only a value of visibility, there is
no need to merge rankings. What matters for the
evaluation of visibility on the Web are just the
placements in the various search engines of the or-
ganization official website that we are calculating
Smart Web Visibility. In the evaluation of web visi-
bility, the metasearch engine is used as follows: the
WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies
672

name or acronym of the organization you want is
consulted at n search engines; then the official web-
site of the organization (previously known) is lo-
cated in each of the n classifications, featuring a
second moment of searching; finally the visibility of
the organization is scored according to the official
website placement in the search pages rankings.
The evaluation method scores the placement of
webpages in search engines as the classic method
Borda Count (Saari, 1985), used in voting processes
(Black, 1976) and also widely applied to computa-
tional problems such as rankings fusion and meta-
search (Aslam and Montague, 2001). For making
use directly of the placement of the elements of a
ranking as means of scoring, the method satisfies the
needs of the proposed evaluation way of web visibil-
ity. Borda Count is, basically, an election method in
which each elector ranks the candidates in order of
preference. The winner is determined by points giv-
en to each candidate according to the position they
are in the list of preference of each voter. The candi-
date with the highest score at the end of the count is
the winner. To determine the score for each place-
ment, we need to know the number of candidates.
Thus, for n candidates, the first receive n points, the
second n-1, and so on. Alternatively, we can assign
one point for the first place, 1/2 for the second, 1/3
to third, and so on, giving emphasis on the first
place. Table 1 below illustrates the first scoring me-
thod, for n = 5.
Table 1: Scoring by Borda Count.
Placement Formula Score
1º n 5points
2º (n-1) 4points
3º (n-2) 3points
4º (n-3) 2points
5º (n-4) 1 point
As the elements that will be voted in the context
of this work are webpages, we can’t know all the
candidates. The voters are search engines and they
tend to rank a varying quantity of webpages. Also,
the rankings do not necessarily contain the same
webpages.
The solution for the problem is to use only the
first places of each search engine, known as top-n.
As the rankings will be used for evaluate visibility
and it is known that users of search engines tend to
focus only on the first places, truncating the results
should not affect the results nor distort the value cal-
culated. According to our previous studies, work on-
ly with top-10 of the search engines is ideal, because
as we increase the value of n, the greater is the noise
in the rankings, i.e., increases the number of un-
wanted sites (irrelevant) returned by the query. Thus,
using the top-10, the first placed website receives ten
points, the second nine, until the tenth that receive
just one point.
3.1 Evaluating Web Visibility
For a given organization to which we want to meas-
ure the visibility on the web, the evaluation of web
visibility helped by metasearch and based on the po-
sitions of the official website of the organization in
the various rankings proceeds as follows: i) Identifi-
cation of the organization official website; ii) Search
by acronym (or name) of the organization in n
search engines; iii) Search for the official website in
each of the n rankings; iv) Sum of scores according
to the position in each ranking.
We have implemented a prototype for experi-
mental purposes, so we can efficiently produce rank-
ings of institutions belonging to the same domain. In
the prototype were included fourteen search engines.
The input parameters are the target of the search and
the website. The query is sent to fourteen search en-
gines and the top-10 of each ranking are placed in a
matrix where each column represents a search en-
gine and each row represents a retrieved webpage.
As among the top-10 of each search engine does not
necessarily appear the same ten pages, a zero value
is assigned to cells in the matrix that correspond to
pages that did not appear in the top-10 search engine
column. In the other cells of the matrix are assigned
the placement of the websites according to the rank-
ings. Of this matrix is utilized only the row corres-
ponding to the organization's official website. For
each non-zero value of the line are assigned and
summed points, according to the Borda Count,
reaching a maximum value of 140, in which case the
official website returned in first place in all fourteen
search engines. The search engines involved are:
Brazilian versions of Alta Vista, Ask, Google and
Yahoo, global versions of Alexa, All The Web, Alta
Vista, Ask, AOL, Exalead, Google, Icerocket, Lycos
and Yahoo. The choice of search engines will affect
directly the results, so this is a very important step.
A problem that can occur with this way of mea-
suring visibility is when there are other organiza-
tions using the same name or acronym. This reduces
the visibility value calculated and characterizes the
problem of homonyms, since the organizations will
compete for positions in the same query in the
search engines. A way to eliminate the problem is
specifying the domain, adding semantics to query in
the case with the query expansion including academ-
ic terms.
SMARTWEBVISIBILITYOFORGANIZATIONS
673

4 UNIVERSITIES RANKING
Universities characterize a homogeneous domain,
where each institution has an acronym and a web-
page that does not vary very much from certain
standard, being an ideal study case for ranking or-
ganizations based on the vision of their official web-
sites by search engines. The use of information from
the Web to rank universities is nothing new. The QS
World University Rankings, by QS Quacquarelli
Symonds Limited, uses Scopus, which is a database
(available in Web version) of abstracts and citations
of scientific literature production, to measure the in-
tensity of research through the documents recovera-
ble by the platform. The Academic Ranking of
World Universities (ARWU), by Shanghai Ranking
Consultancy, uses data sources of the Web to define
their classification criteria which involve number of
publications, citations and awards received by the
researchers. Another ranking of universities world-
wide is the Webometrics Ranking of World Univer-
sities, by Cybermetrics Lab, which uses Webome-
trics and its sub-areas, particularly the Web Visibili-
ty. In its rankings, the Web Visibility represents
50% of the total score aggregate to the university,
and this visibility is measured by Yahoo! Search,
taking into account the total number of unique in-
links that each official university’s website receives.
There are several rankings classifying higher
education institutions worldwide on the web, as can
be seen in the work of the Nordic research council
Nordforsk (2011). The analysis of the presence of
the universities by means of cybermetric indicators
shows up as an important tool for evaluations and
comparisons, being increasingly more relevant. A
good placement within a ranking can attract more
high-level researchers, students and investment for
the university. Universities have become aware of
the importance of their presence on the web. A way
to maximize the visibility of an institution is to
maintain a digital repository that represents the
scientific output of the institution (Swan and Carr,
2008).
The following section is the applying of the for-
mula developed by this work to the specific domain
of the Brazilian universities.
4.1 Brazilian Universities Ranking
Thirty Brazilian universities were submitted to the
Smart Web Visibility evaluation. The universities
were chosen based on the set of Brazilian universi-
ties of the Webometrics Ranking of World Universi-
ties for future comparisons. The acronym of each
university was used as a query parameter in the pro-
totype developed, along with their respective official
websites previously identified, revealing the visibili-
ty of the acronym linked to the university in the web.
Table 2 contains the top-10 universities sampled in
the experiment, 140 being the maximum visibility
value, corresponding to fourteen first places.
In the ranking of table 2, containing the fifteen
best placed universities among the thirty submitted
to the Web Visibility evaluation, there is the Pontifi-
cal Catholic University of São Paulo (PUC-SP) as
leader and the one with maximum points. After is
the Pontifical Catholic University of Rio de Janei-
ro(PUC-Rio) showing that both acronyms have a
great power of discrimination and excellent visibility
in the eyes of search engines. The Federal University
of São Paulo (UNIFESP) completes the top-3, fol-
lowed by five universities tied in fourth place, two
tied in ninth place, UNICAMP in eleventh, three
universities again tied in twelfth and UFPR complet-
ing the top-15.
Table 2: Top-15 sampled universities.
Rank University Score
1º PUC-SP 140
2º PUC-Rio 138
3º UNIFESP 137
4º UFRGS 135
5º PUCRS 135
6º UFRN 135
7º UFSM 135
8º UERJ 135
9º UFRJ 134
10º UFSCAR 134
11º UNICAMP 133
12º UFPE 131
13º UFPB 131
14º UNISINOS 131
15º UFPR 130
It is interesting to compare the results with the
ranking of the Brazilian institutions on the Webome-
trics Ranking of World Universities. Of the top-3
ranking, USP and UNICAMP (1st and 2nd, respec-
tively), did not make the top-10 in this experiment.
Considering that they are two major universities in
Brazil, and also that among the webpages returned
by metasearch featured many that are unrelated to
universities, a new form of querying was expe-
rienced. The homonymy problem was evident in this
first experiment. Some universities were affected, as
the Federal University of Ceará (UFC) whose
acronym also belongs to an organization most fam-
ous, the Ultimate Fighting Championship. In this
first ranking the UFC was in the thirty position. In a
WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies
674

new experiment, the acronyms of the same thirty
universities were resubmitted to the query adding the
word 'university'. This process is known as Query
Expansion, and serves to define more precisely what
we are looking for, in this case universities, avoiding
the homonymy problem. The new ranking is shown
at table 3.
Table 3: Top-15 sampled universities with query exten-
sion.
Rank University Score
1º UNICAMP 140
2º UEM 140
3º UFV 140
4º USP 136
5º UFRGS 136
6º UNB 136
7º PUC-Rio 136
8º UFPE 136
9º UFF 136
10º UFRN 136
11º PUC-SP 136
12º UFPB 136
13º UNESP 135
14º UFSC 135
15º UFC 135
Now, in the first place was a tie of three universi-
ties, which were not ranked in the top-10 before, and
they obtained maximum score. UNICAMP, together
with the State University of Maringá (UEM) and the
Federal University of Viçosa (UFV) had a score of
133, 77 and 83 in the previous experiment, respec-
tively, obtaining the maximum score in the version
with the expanded query. Next, nine universities
were tied with 136 points, among them the PUC-SP
and PUC-Rio, showing that the technique has in-
creased the noise level in the rankings by search en-
gines, which already had a great precision for these
acronyms, but not impaired, as they continue well
placed. Also in the ranking of table 3, there is a good
placement of the USP, which rose from 63 to 136
points, entering the top-10.
In general, the word 'university' with the
acronym in the query raised the score of all universi-
ties sampled. In the new version of the experiment,
the university with less scoring scored 90 points, in
contrast to 48 points in the first experiment. The to-
tal score of all the thirty universities, using the query
only by the acronym were summed 3466 points, us-
ing the expanded query the sum reached 3910
points. The graphic of figure 1 shows the increasing
in scores for some of the most prestigious universi-
ties of Brazil.
Figure 1: Scoring with and without the homonymous prob-
lem.
The homonymous problem of universities like
UFC was avoided by Query Expansion and allowed
the university to rise from the thirty position to the
tenth fifth position. Of course, if other organization
linked to universities has the same acronym the
problem would not be solved by simple adding the
word ‘university’ in the query. A more sophisticated
query expansion would be necessary to include extra
semantics.
5 CONCLUSIONS
As seen, the visibility of an organization on the Web
can be measured in several ways. The most common
form is the count of the unique external inlinks, as
used by Webometrics Ranking. Another way to
measure the presence on the network can be
represented by the number of webpages recovered
from search engines or articles indexed. The univer-
sities rankings of the Internet use mostly bibliomet-
ric indicators, especially citation.
The main contribution of this work is a new way
to evaluate visibility on the web, an indicator based
on data from search engines. When performing a
search on any search engine, users tend to look only
at the first results. That is, a website well placed in a
search engine has, in other words, a good visibility
in such search engine. This idea was explored: the
placement of the website in various search engines,
rather than the number of pages on the domain of the
institution, number of documents, citations or links.
It was presented a way of evaluate visibility on the
Web through search engines, taking into account the
placements of the webpage linked to the search ar-
gument in several rankings. The evaluation way pre-
sented was named Smart Web Visibility and shows
how well a particular entity is perceived by web
search engines.
SMARTWEBVISIBILITYOFORGANIZATIONS
675

Through a study case in order to rank universities
by Smart Web Visibility, we observed an interesting
application of the evaluation proposed, showing a
current scenario that is the subject of several re-
searches. Applying metasearch on the universi-
ties’acronyms, two rankings were developed: one
showing the visibility of webpages of institutions
when a search made with only the acronym, and
another using a query expansion technique to better
describe the domain, increasing the scoring of the
universities sampled in the experiments and avoid-
ing the homonymous problem.
The Smart Web Visibility has applicability in
any field, not only universities, but for the genera-
tion of rankings is important that the domain is ho-
mogeneous. Future studies should seek a way to
demonstrate the ampleness of the method.
5.1 Future Work
As mentioned above, efforts are still required to
prove the application of Smart Web Visibility evalu-
ation generically, allowing us to develop rankings in
other domains. Furthermore, the work identified the
possibility of some future studies like the study of
other parameters that can be extracted for the eval-
uation of web visibility, the study of tiebreakers for
visibility rankings, and the study of a distribution of
different weights to each search engine according to
some criterion to be studied too. Future works will
be concerned about two main topics. One of them is
to add more semantics to the description of the do-
main, perhaps by ontologies, making possible to na-
vigate through the domain levels. The other main
topic is about extracting time and spatial data with
the metasearch, aiming to discover where and when
the visibility of the target was better or worst. In the
near future, rankings with more universities, includ-
ing universities outside of Brazil, should be devel-
oped.
ACKNOWLEDGEMENTS
This work has been partially supported by CNPq and
by CAPES, Brazil.
REFERENCES
Aaltojärvi, I., Arminen, I., Auranen, O. and Pasanen, H-
M.(2008). Scientific Productivity, Web Visibility and
Citation Patterns in Sixteen Nordic Sociology De-
partments. ActaSociologica, 51(1), 5-22.
Aguillo, I. F.,Granadino, B., Ortega, J. L. and Prieto J. A.
(2006a). Scientific Research Activity and Communica-
tion Measured with Cybermetrics Indicators. Journal
of the American Society for Information Science and
Technology, 57, 1296-1302.
Aguillo, I. F., Granadino, B., Ortega, J. L. (2006b). Brazil
Academic Webuniverse Revisited: A Cybermetric
Analysis. In Proceedings… International Workshop
on Webometrics, In-formetrics and Scientometrics&
Seventh COLLNET Meeting. Nancy, France.
Aguillo, U. F. and Kretschmer, H. (2004).Visibility of
Collaboration on the Web. Scientometrics, 61, 405-
426.
Aslam, J. A. and Montague, M. (2001).Models for Meta-
search.In Proceedings… ACM SIGIR Conference on
Research and Development in Information Retriev-
al.SIGIR'01. ACM, New York, NY, 276-284.
Barjak, F. and Thelwall, M. (2008). A Statistical Analysis
of The Web Presences of European Life Sciences Re-
search Teams. Journal of the American Society for In-
formation Science and Technology, 59, 628-643.
Björneborn, L. and Ingwersen, P. (2004).Toward a Basic
Frame-work for Webometrics. Journal of the Ameri-
can Society for Information Science and Technology,
55, 1216-1227.
Black, D. (1976). Partial Justification of the Borda Count.
Public Choice, 28(1), 1-15.
Cubestat. (2008).Cubestat: The Free Website Value Cal-
culator. Retrieved in November 21, 2011, from http://
www.cubestat.com
Dnscoop. (2009). Domain and SiteValueTool.Retrieved in
November 21, 2011, from http://www.dnscoop.com
Espadas, J.,Calero, C. andPiattini, M. (2008). Web Site
Visibility Evaluation. Journal of the American Society
for Information Science and Technology, 59, 1727-
1742.
Gori, M. and Witten, I. (2005). The Bubble of Web Visi-
bility. Commun... ACM, 48, 115-117.
Kretschmer, H., Kretschmer, U., Kretschmer, T. (2007).
Reflection of Co-Authorship Networks in the Web:
Web Hyperlinks Versus Web Visibility Rates. Scien-
tometrics, 70, 519-540.
Nordforsk.(2011). Comparing Research at Nordic Univer-
sities using Bibliometric Indicators. NORIA-net. Re-
trieved in August29, 2011, from http://www.nord
forsk.org/files/rapp.bib.2011.pub_21.5.11.
Page, L., Brin, S., Motwani, R., and Winograd, T. (1999).
The PageRank Citation Ranking: Bringing Order to
the Web. Technical Report. Stanford InfoLab.
Saari, D. G. (1985). The Optimal Ranking Method is the
Borda Count. Discussion Papers, (638). Northwestern
University.
Swan, A. and Carr, L. (2008) Institutions, their Reposito-
ries and the Web. Serials Review, 34, 31-35.
Viegas, F. B. (n.d.). Word Tree. Retrieved in November
13, 2011, from http://fernandaviegas.com/wordtree.
htm
WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies
676