NEW PERSPECTIVES FOR SEARCH IN SOCIAL NETWORKS
A Challenge for Inclusion
Júlio Cesar dos Reis
Institute of Computing (UNICAMP) and CTI Renato Archer
Rodovia Dom Pedro I, km 143,6, 13069-901, Campinas, São Paulo, Brazil
Rodrigo Bonacin
CTI Renato Archer and Faculty of Campo Limpo Paulista (FACCAMP)
Rua Guatemala, nº 167, 13231-230, Campo Limpo Paulista, São Paulo, Brazil
M. Cecilia C. Baranauskas
Institute of Computing at University of Campinas (UNICAMP)
Caixa Postal 6176, 13083 970, Campinas, São Paulo, Brazil
Keywords: Semantic Web, Semantic Search, Inclusive Social Network Services, Organisational Semiotics.
Abstract: The world is populated with many scenarios characterized by a diversity of cultures and social problems.
Thus it is necessary to investigate computational solutions that respect this diversity. The use of search
engines is one of the main mechanisms to provide the access to information generated in the Social Network
Services (SNS). These mechanisms are currently built through lexical-syntactical processing resulting in
barriers for many users to access correct and valuable information in the Web. Novel search mechanisms
could effectively help people to recover and use information through Inclusive Social Networks Services
(ISN), promoting the universal access to information. This paper shows results of search activities in an ISN
that point out how to improve search engines considering aspects related to social and digital inclusion.
Inspired in these results, we outline an approach based on Organisational Semiotics to build Web ontology,
which is used by an inclusive search engine drawn up in this paper. Actually, this proposal combines
different strategies to provide better search results for all.
1 INTRODUCTION
Web systems and portals are available to a vast
number of people with socio-cultural differences.
Within a Social Network, people communicate and
behave according to commitments, linguistic
constraints, culture and other social aspects. Social
Network Services (SNS) of nowadays could be more
appropriate to the context of people´s life,
considering their differences, in order to promote
social and digital inclusion.
This scenario becomes even more challenging
and important in contexts such as Brazil, where there
are several social problems and a huge cultural
diversity. In this perspective, one of the fundamental
points is to provide barriers free access to
information to every citizen. This could help to bring
great social benefits and contribute to a profound
social transformation. Thus, it is extremely
important that all people have the opportunity to
retrieve, access and use information provided in the
digital media in a smooth way.
The SNS represents an opportunity to
interaction, access to information and knowledge
through the Web. These systems primarily allow
individuals to share their interests and activities,
constituting communities. The e-Cidadania project
(Baranauskas, 2007) aims at transforming a SNS
into an engine for digital inclusion and citizenship.
The networks with such characteristics can be
defined as "Inclusive Social Network Services”
(ISN) (see Neris et al., 2009). The use of search
53
Reis J., Bonacin R. and C. Baranauskas M. (2010).
NEW PERSPECTIVES FOR SEARCH IN SOCIAL NETWORKS - A Challenge for Inclusion.
In Proceedings of the 12th International Conference on Enterprise Information Systems - Human-Computer Interaction, pages 53-62
DOI: 10.5220/0002897000530062
Copyright
c
SciTePress
engines is one of the primary ways to find and to
make access to information generated in these
systems. However, search mechanisms are currently
built based on comparisons of keywords and lexical-
syntactical information processing (syntax search).
These mechanisms are not sufficient and adequate to
effectively make sense to individuals in an inclusive
scenario within social networks.
Based on empirical results, which will be
discussed in this paper, we have observed that
people organized into virtual communities bring to
this space their own vocabularies and meanings, and
also develop their own local vocabularies through
interaction and communication using technology.
The results pointed out the need for novel search
mechanisms considering the diversity of users’
competencies and inclusion aspects.
A more appropriate inclusive search solution to
an ISN should reflect the semantics used by
participants of the social network. In few words, a
search engine should take into account the local
meanings created, shared and used by people
organized into a community. In this paper we argue
that the quality and response accuracy of a search
mechanism are intrinsically associated to the
proximity of the semantics shared by people. Thus,
it is necessary to identify the meanings used in the
network and to represent its semantic aspects. This
could actually contribute to make the information
accessible to everyone, including people with low
educational levels that have difficulty to access
online information due to their simple vocabulary or
their deficiency in writing. Usually, these people use
an informal (colloquial) or local vocabulary in the
search. With the proposed solution they could find
the correct information in an easier and more precise
way, and learn from it.
In this paper we show results of search activities
within an ISN, conducted in the context of the e-
Cidadania project. The goal of these activities was to
observe a set of search scenarios with potential users
of an ISN, and to understand how these users make
sense of a search mechanism. Based on the results,
we present a proposal of developing a more
appropriate search mechanism for an ISN with
foundations in Organisacional Semiotics (OS)
(Stamper 2001; Liu 2000). In our approach the goal
is to expand and to improve the search technologies
and techniques of the Semantic Web field. Besides,
the representational structure (semantic model) used
by the search mechanism is based on data from the
interaction and communication of users in the social
network system. Thus, the search engine will take
into account the meanings shared and created by
people (including the informal terms) in their
interaction with the system aiming to provide better
results.
The paper is organized as follows: Section 2
presents the concept of ISN and the importance of
search mechanisms for the universal access to
information; Section 3 presents the analysis of the
empirical experiment with ISN users; Section 4
details the proposed approach; Section 5 makes a
discussion about the approach and related works;
and Section 6 concludes presenting further works.
2 UNIVERSAL ACCESS
AND INCLUSIVE SEARCH
According to Boyd & Ellison (2008) since the
beginning of the Social Networking, sites such as
MySpace, Facebook, Orkut and others have attracted
millions of users and many of them have integrated
these sites into their daily practices.
Online Social Networks or "communities of
members" have great relevance in the Web as users
spend much time of navigation on them. According
to Nielsen (2009), social networks are more popular
than e-mail, with 66.8% of global reach. Around the
world it represents the fourth most used resource in
the Internet and 85.2% of penetration are in the
portals and communities of general interest.
Additionally, 85.9% of Internet users use search
engines, which is one of the most common activities.
Despite these great numbers and the success of
Social Network sites among Internet users, in social
contexts like Brazil and other developing countries,
there are yet a lot of people without access to the
Internet and consequently without opportunities to
access information and knowledge. Social indicators
shown by the PNAD (National Survey by Household
Sample) produced by IBGE (Brazilian Institute of
Geography and Statistics) (IBGE, 2008) points out
that in 2008, 65% of the population did not have
access to the Internet.
In addition, important data from the Ministry of
Education in Brazil (MEC, 2007) reveal that about
30 million of Brazilians are functionally illiterate,
defined as the population over 15 years old and less
than 4 years of schooling (21.6% of the population).
Using a broader concept of functional illiteracy,
according to a survey from Paulo Montenegro held
in 2007 (IPM, 2007), the majority (64%) of
Brazilians between 15 and 64 years old and more
than 4 years of schooling reach no more than the
degree of rudimentary literacy, i.e., they have only
ICEIS 2010 - 12th International Conference on Enterprise Information Systems
54
the ability to locate explicit information in short
texts or make simple math, they are not able to
understand longer texts. This data illustrates only
part of the challenge that we face in terms of
designing systems, which should include all these
users. In this context it has become a major concern
to allow access to online content available from SNS
to all people in a more "natural" and efficient way.
Thus, it is extremely important to recreate
methods to permit the effective access and use of
information conveyed in digital media, for all. This
could be materialized with the ISN concept. We
understand ISN as a “virtual communication space”
based on the concept of social networks, which is
inclusive and allows the community to share
knowledge about the community know-how. This
space has to facilitate “exchange” (of knowledge,
goods and services) in accordance to the
collaborative (project team, partners, community)
system conception.
It is also worth to mention that in an ISN there
are not target users, but all users are relevant and
should be included without discrimination. Thus,
there may be people without skills to handle certain
technological features of the system and
consequently without knowledge to find information
that they need in the system. Moreover, those users
most often use colloquial terms to express
themselves through the system. For example, they
may use the term "postinho" (in Portuguese) instead
of "Basic Health Unit" (formal). They use terms that
make sense to them, but in fact, these expressions
semantically mean the same. So when someone is
trying to retrieve information from the ISN, these
factors must be taken into account by the search
engine. On the other hand, when a user searches for
something in a non-formal or not refined way and,
the same concept but in its formal way (cult) is
returned, this represents an opportunity for learning.
Therefore, we should seek for a computational
search solution that takes into consideration the
meaning that is adopted or emerges in the context of
use of that network; i.e. the meaning that people
bring to the network, and those that are constructed
by using the system over time (through interaction).
This may facilitate and provide better access to the
content generated by users of the network.
3 ANALYSING SEARCH
SCENARIOS OF AN ISN
From a practical point of view, the e-Cidadania pro-
ject resulted in the ‘Vilanarede’
(www.vilanarede.org.br) ISN system. This system
has represented an opportunity to investigate the
interaction behavior of representative users in a
developing country. As a direct activity of the
project, we have conducted the 8th Participatory
Workshop, in a telecenter located at ‘Vila União’,
neighborhood of the Campinas, Brazil. In this
workshop we developed an activity related to search
in the ISN. The objective was to observe some major
points including: (1) How would the users build
understanding of the search engine? (2) Which key-
words would they use? (3) Would they have any
difficulty in completing the proposed scenarios? and
(3) What would be their satisfaction with the search
results?
A task sheet with 4 search scenarios was
presented to each pair of participants, and a form
was prepared to the observers (researchers) of the
activity. We had 7 pairs of users in total. An initial
instruction about the activity was given to the
participants. The pairs were formed by the users
themselves, and for each scenario the pair should
write the words used in the search and the title of the
announcements found. Resulting from this activity,
we had both the sheet tasks filled by the pairs of
users and the observation forms filled out by the
observers. Besides, the activity was filmed and there
was audio recording of each pair during the task
execution. The 4 search scenarios were:
Scenario 1: Find out announcements on how to
popularize the ‘VilanaRede’.
Scenario 2: Find out announcements of mango
(fruit) in ‘VilanaRede’.
Scenario 3: Find out announcements related to food
in‘VilanaRede’.
Scenario 4: Find out announcement related to
religion item combined with handicraft in
‘VilanaRede’.
Each scenario intended to verify whether
semantic capacity was needed for the search
mechanism. The time for the completion of the
scenarios was approximately 45 minutes. After the
execution of the search scenarios, a general
discussion was conducted to get the general
impression of users about the activity. During this
discussion, several interesting stories were collected.
In Scenario 1, we wanted to observe whether
users would use synonymous of “popularize” to find
the announcements. Some pairs had difficulty to
understand the scenario, as well as difficulty in
choosing the terms for the search. However some
pairs associated the word "popularize" to "divulge"
NEW PERSPECTIVES FOR SEARCH IN SOCIAL NETWORKS - A Challenge for Inclusion
55
and quickly found related announcements. In this
scenario one pair used some unusual keywords such
as: "boca-a-boca” (a popular expression used in
Brazil that means “orally passing information from
person to person”), “email”, “phone” and “posters”.
By using the term "boca-a-boca" in order to find
announcements about how to divulge the site,
unusual results also appeared as an advertisement for
"Bife de casca de banana” (steak of banana peel). It
happened because in one of the comments of this
announcement we find "I'm with water in my mouth
(boca)” in reference to the announcement of "steak
of banana peel”. Phrases for search like "divulgation
of the ‘Vilanarede’" or verbs such as "to popularize
announcement" or "advertising Vila” were also used
in this scenario.
In Scenario 2, we wanted to verify if users would
find any announcement related to mango (fruit) in
the application. There was no announcement about
the mango fruit in the system. However there was an
announcement about mangá (cartoon) and it was
written without the acute accent in the word
(‘manga’ in Portuguese, which is mango fruit in
English). In this scenario, users mainly used the
keywords: "mango (fruit)", "fruit", "mango",
"mango fruit", "mango / fruit". Some pairs were
uncertain if they would have to put the keyword
"fruit" or not. Note that in a semantic search, by
putting the key-word “fruit”, the application should
return all the announcements with mango (fruit), in
the case of announcements semantically related to
fruit.
In Scenario 3, we wanted to see whether users
would use the keyword "food" in the search or they
would make a search for specific foods through the
search engine. As a result, when users tried the
keyword "food", the system returned nothing.
However there are several announcements on food
in the system: the sale of “salgadinhos” (homemade
snacks), “pão de queijo” (cheese bread) and others.
Among the relevant considerations from the
observers, during the execution of this scenario users
said that the system should relate “salgadinhos
(homemade snacks), “pão de queijo” (cheese bread)
and “Bife de casca de banana” (steak of banana
peel) with the concept of food. And this makes sense
because semantically all of these are food. During
the discussion phase one of the users commented:
"Using food is easier because it already covers
everything," i.e., all types of food in the system.
Another said: "To be more 'lean' and practical for
those who are starting (in terms of computer
literacy), like us, when we enter “food”, it should
return a variety of foods due to our difficulty." Yet
another user says: "Maybe to use food does not help
in the search of something more specific, but if it is
something that we have no knowledge of the
domain, or we do not know what to look for, the tool
would be useful and helpful." The main keywords
used in this scenario were "food", "comida caseira"
(homemade cooking), "food sale", "salgado"
(homemade snacks), "salgadinhos"(small homemade
snacks), "salgadinho frito" (fried homemade
snacks), pies, "doces" (sweets), "pão-de-queijo
"
(cheese bread), "docinhos” (small sweets), cake,
pastel and “brigadeiro” (chocolate sweet). Note that
users utilize several variations in words such as
“homemade snack”, "small homemade snack" and
"fried homemade snacks".
With Scenario 4, we aimed to determine which
keywords users would use when looking for a
specific announcement. One of the observers
indicated that the pair found the "Saint Anthony"
because they already knew that this announcement
was in the system. The same was reported by several
other observers. The vast majority of the pairs used
the keywords: "homemade craft", "Crafts saint",
"holiness", "holy" and "saints". Users found the
desired information successfully. But one of the
pairs put keywords like: "Orisha", "Orisha of cloth",
"religious" and "sculpture" and didn’t find out any
announcement. Several observers noticed that the
subjects utilize terms from their own colloquial
language in the search; examples can be seen as
"manga rosa" (pink mango), "manga coquinho"
(coconuts mango), "tutu de feijão" (tutu bean),
"boca-a-boca" for the term “divulge”, “small
sweets”, "little homemade snack" and "Orisha". Also
in several occasions the pairs discussed before
reaching an agreement on which word to use in the
search.
Another interesting result was obtained by the
analyze of the interaction of a deaf-mute user with
search mechanisms. We could observe that this user
has some difficult related to word meanings. We
observed that he does not know some words, and use
the same hand signal to several different words; he
has difficult in writing and make different meaning
associations, because what make sense for them
(many deaf-mute people) is not words, but signals.
The user had difficult in understanding the scenario
1, since the words popularize, advertise,
advertisement and disclose have the same hand
signal in his language. Moreover we could see that
the user understands different some words that has
the same meaning; his behavior during the search
was not confident neither independent on
performing the tasks, and the user asked a lot of
ICEIS 2010 - 12th International Conference on Enterprise Information Systems
56
questions to the observer.
Additionally, general results indicate that users
from this context under study (prospective users of
an ISN) had difficulty with the search button; in
other words, they do not have a clear concept about
the act of "searching" in an application on the
Internet. Some users had no idea about the scope of
searching. They do not know if that search is only
for the announcements in the ‘Vilanarede’ system
which they are using or whatever the term they type
the system will return successful results. This stay
explicit in a description from a user who said:
"Search fondée because it is something chic”.
However another report from other user says:
"Fondée is very chic, we do not have it here in our
network... we will not put it in the search because
the network is ours, it is “poor”... and it will not
have fondée...". This statement shows that the
second user has the notion of the search scope,
which will be just within the announcements from
that social network system, so as there were any
announcements about fondée, nothing would be
returned.
Even with this lack of sense about the search
scope, one of the observers explains that users stay
surprised with the power of the search, and they
explored and tested it easily. This is confirmed due
to the users’ behavior during the execution of the
activity. Positive results were obtained from it. From
the data filled by the observers’ formularies, it is
possible to indicate that approximately 80% of the
pairs proved be safe during the performance of the
task. Approximately 80% of the pairs proved be
independently by performing the scenarios. Around
60% of the pairs did not make a lot of questions to
the facilitator during the task. In the discussion after
the activity, users said that they had to reason more
to perform the search, making exchange of words
and testing various alternatives. More details from
this experiment can be seen in Reis et al. (2010a).
These practical results show that users
colloquial language should be considered during the
development of more appropriate search engines.
Moreover, people in a social network can create
their vocabulary, sharing meanings in the
community. The results showed us that it is
necessary to construct computationally tractable
models from the semantic point of view that come
out from the network itself. Semantics here is
understood as the interpretation of signs (Peirce
1931-1958, cf 2.228) by individuals and their
association with real world elements. This
interpretation is socially contextualized; i.e.
individuals and communities may have different
interpretations for the same sign and a sign may
connote different meanings depending on the
context applied.
As a theoretical reference for our proposed
approach, we have used the Semantic Analysis
Method (SAM) from the Organisational Semiotics
(Liu, 2000). This method assists the users or
problem owners in eliciting and representing
meanings in a formal and precise model. An
Ontology Chart (OC) describes a view of
responsible agents in the focal domain and their
behavior or action patterns named affordances (Liu,
2000). Some basic concepts of SAM adopted in this
paper are based in Liu (2000). In the next section we
outline a new approach to build Web ontologies
based on this method as well as a novel search
mechanism that utilizes it.
4 TOWARD A NEW SEARCH
MECHANISM
General difficulties faced by users to get information
in the Web can be mainly explained by: (1) overload
of information presented in the system; and (2)
problems related to the contextualization of meaning
for the terms used. As an attempt to solve this
problem, we have investigated approaches that can
result in better and more appropriate search engines
for ISN.
In a social network the “emergence” of meaning
is an ongoing process in which meanings and
interpretations are constructed, used and shared
through the system based on the interactions and
expressions of users. These interpretations expressed
by users in the system could be computationally
represented. Several improvements could be
achieved such as semantic models that can make the
social network context more faithful resulting in
more adequate search engines.
To accomplish that, we propose a search engine
informed by a Semiotic approach. We have
developed a semi-automatic process to model the
semantics in the ISN using SAM, and we use the
outcome of this process in the search engine.
4.1 Modelling the Ontologies for ISN
In the ‘VilanaRede’ system, users express
themselves through their profiles, announcements of
products, services and ideas created by them; and
they communicate mainly through commentaries
about the announcements and chats between
NEW PERSPECTIVES FOR SEARCH IN SOCIAL NETWORKS - A Challenge for Inclusion
57
members of the network. These data are stored in the
ISN system database and from these dada we
represent the semantics used in the social network in
a structure called ‘Semiotic Web ontology’.
This structure is a semantic model
(computationally tractable ontology) constructed
from a semi-automatic process based on the SAM
along with the vocabularies shared in the social
network. The theoretical and methodological
concepts described in the SAM are used in
conjunction with other technologies from the
Semantic Web field to describe computationally
tractable ontologies using the Web Ontology
Language (OWL) (W3C, 2004). The idea is to
incorporate the concepts of particular Agents (roles)
and Affordances (pattern of behavior) arising from
the SAM into an expanded and more representative
Semantic Web ontology. It is worth to mention that
it is not the goal to create a “perfect ontology” from
a theoretical point of view, but to produce practical
and immediate results for search in ISN. Therefore
some properties from the OC may not be fully
transcribed to OWL, while other aspects such as
agent-affordance relationship are emphasized.
This approach is justified from a Semiotic
perspective, since the signs are socially constructed.
Thus, a computational model that represents the
semantics of an SNS should contain the agents that
interpret the socially shared concepts. With this
approach we incorporate and take into account to
Semantic Web ontologies concerns and possible
representations arising from the Ontology in a
Semiotic perspective. In addition to agents and
affordances, we have observed that Semantic Web
ontologies also do not incorporate (at least
explicitly) the idea of ontological dependency
relations, the existential relation in the model. The
approach is also justified by the representational
limitations shown in literature (e.g. Tanasescu &
Streibel (2007)) regarding the use of ontologies in
computing and their expressivity.
As a conceptual model, in this computational
approach based on SAM, the agents have
behavior(s) (affordances) related to a concept. For
example, a seamstress, which is an agent, can sew a
manga” (it means sleeve in English). Sewing is a
pattern of behavior of a seamstress (in other words
an affordance). “Manga” is a concept that can have
several different meanings in Portuguese (It can
mean sleeve, fruit, color, etc.), but in this context
due to the affordance and the agent, the meaning of
manga” is more closely linked to shirt and not, for
example, to “manga” fruit (mango in English) that
can also be represented in the model, as illustrated
by Figure 1.
Figure 1: Modelling meanings: an example of polysemy
using agents and affordances.
Figure 1 illustrates an example of modeling in
which the grocer and the seamstress are agents that
have affordances connected to specific concepts.
And this model can also have specific ‘is-a’
relationships; for example ‘manga rosa’ is a specific
kind of mango. This also shows that concepts can be
related to several agents and affordances and with
other concepts, constituting relations and
representations that make more complete ontologies
compared to conventional ontologies described
purely for a domain. For example, ‘manga’ can also
mean a color for a painter who is searching
something in the network, as well as ‘manga’ can
have any synonym that makes sense for an agent ‘Y’
modeled from the data of the social network. We can
see other examples like: ‘crane’ mean a bird or a
type of construction equipment and we can model it
using the agents and their affordances in a ‘Semiotic
Web Ontology’.
To develop this representation, we propose an
assisted method (semi-automatic) with several
distinct steps, which is a novel approach to create
ontologies; the method is illustrated in Figure 2. It
includes: (1) the extraction of terms from the
database of the ISN system; (2) the creation of an
OC (from SAM); and (3) the creation of the final
OWL ontology.
In this assisted process, the first step is to process
the data from the system database. This step takes
into account the social relations in the network, and
provides the necessary well defined data (a list of
concepts, agents and affordances, etc) to build the
semantic model. Related to this phase, we apply
algorithms for text extraction and analysis and data
from the database of the ISN system. For this we
propose the use of keyphases extractors like KEA
(Medelyan, 2006) and tools for term extraction and
creation of concepts such OntoLP (Ribeiro & Vieira,
2008).
ICEIS 2010 - 12th International Conference on Enterprise Information Systems
58
Figure 2: Illustration of the Semi-Automatic Tool.
The next step is the building of an OC (from
SAM) by an ontology engineer. This intermediate
ontology diagram is important to identify the
possible agents in the ISN and their patterns of
behavior.
In the third step, from the OC, a set of specific
heuristics and rule transformations is applied to
derive an initial OWL ontology (computationally
tractable), extending the computational development
of the SONAR CASE tool (see Santos et al., 2008).
Bonacin et al. (2004) proposed a heuristic to
transform OC into system design diagrams; however
those heuristic must be adapted to our purpose. In
this step the ontology engineer can also be assisted
by existing tools for Ontology Learning and other
data extracted from the system. As suggested by
Maedche & Staab (2001), Ontology Learning can be
helpful for ontology engineers to build these
artifacts. In the next sub-section we draw up an
Inclusive Search mechanism.
4.2 Outlining an Inclusive Search
Engine
After the creation of the semantic model, it is used
by the ISN search engine. When the user is logged in
the social network system and he/she enters with
some search term(s) in the search engine, the system
starts a process of making relationships of this/these
keyword(s) with the available ‘Semiotic Web
Ontology’. The purpose is to perform a search in the
database system and to return semantically related
results. For example, suppose the user types the term
"small snack". If there is nothing in the system with
this expression, from the analogies and semantic
relations made, the system may return some other
types of food semantically close. Likewise if the
user enters the word ‘food’, all advertisements
related to food should be returned.
There are several architectural proposals for
semantic search solutions, as described in Mangold
(2007) and also in Fang et al. (2005). The decisions
and architectural strategies for resolving the
semantic search in this implementation is carried out
in accordance with the requirements of an ISN. The
main difference in the search solution of this
proposal is to take into account information
regarding the user that is making the search (from
his/her profile) and the user that produces the
content. See Figure 3 that illustrates this idea.
Figure 3: Illustration of the Search Mechanism.
In this strategy, the user profile is important due
to the possibility of discovering a context for the
search terms. From the user profile, we aim to
identify the agent represented in the ontology. Thus
we can prioritize (or even limit) the search space,
making a relation between the user and the semantic
model generated. For instance, imagine that a
biologist is logged into the system (we could find
that a user is a biologist based on his/her profile) and
requests a search with the keyword ‘crane’. If there
exist a relation between the ‘biologist’ agent and the
term ‘crane’ in the ontology, most likely the results
(announcements) that could be returned first (ranked
NEW PERSPECTIVES FOR SEARCH IN SOCIAL NETWORKS - A Challenge for Inclusion
59
first) should be related to the concept of crane as a
‘bird’, not the other meaning(s) of this word.
However, to a civil engineer that searches the same
word, maybe the results that most interest him / her
refers to the construction equipment and not the bird.
We do not mean that other results are not required or
may not be returned in response to the search, (the
engineer may want to know about this kind of bird),
but the announcements from the social network that
relates ‘crane’ with a construction equipment must
have greater relevance in the ranking of results.
The agent-affordance relation is also used to
indicate the probable meaning of the words in the
announcement. For example, we could verify
whether the word ‘crane’ is about ‘bird’ or
‘construction equipment’ in a particular
announcement that mentions the word ‘crane’ based
on the user that entered such information, or based
on the commentaries made on this announcement.
Then, if the user who submitted the announcement is
a ‘biologist’ agent, ‘crane’ would be most likely a
‘bird’ in this announcement. In a similar way, if the
advertiser is a civil engineer, also ‘crane’ would
mean most likely a ‘construction equipment’. We
also could have relationships between agents and
could verify how much an agent is semantically
close to another and indicate the probable meaning
based on this aspect.
5 DISCUSSION
The support for better results from the search engine
in this approach demands a careful modeling
procedure. Different signs with the same meaning
(synonyms) coming from different virtual
communities of the social network can be discovered
having the opportunity to be represented in the
ontology; such signs and meanings can be purely
regional. They could not be present in formal
dictionaries or thesauri generally used by
conventional search mechanisms. Furthermore
because they are cultural expressions emerging from
the social network, the ontology would potentially
provide smarter and richer search results when
compared to ontologies based on domains or formal
definitions.
The approach provides means to discover and
distinguish the meanings used in the SNS,
representing them through the agents in the
‘Semiotic Web ontology’. Differently from
conventional computing ontologies, and other
approaches to semantic representation, our proposal
involves adding the agents and affordances concept.
This addition can cooperate for richer search results
treating the polysemy problem in not restricted or
controlled language contexts. Moreover, the
inclusion of the agents and other concepts from
SAM in the Web ontology can help improving the
search mechanism, generating results more adequate
to a SNS context.
In the case of users with limited literacy and with
difficulty in dealing with technological artifacts
(digitally illiterate), it is important to give to them
the opportunity to perform the search using their
daily language since usually is what make sense to
them, and to provide search results more natural and
adequate to their live. Thus, the search engine
should reflect the semantic reality of the social
network users, and therefore the required
information will be easier to find. A search engine
with such characteristics could create opportunities
for inclusion, since the method for building the
semantic model as well as the strategies in the search
engine to use the ontology suggests that the search
results returned will tend to make more sense for the
user that searches.
Some recent studies in the literature address
search solutions for SNS (e.g. Vieira et al. (2007),
Perisic & Haynes (2009) and Gürsel & Sen (2009)).
These works are particularly focused on searching
just users’ profile in the network and to develop
solutions in this direction; the work of Choudhari et
al. (2008) makes progress in the development of
semantic search in social networks, however their
work have the same limitation and does not use
ontologies to perform the search.
Regarding semantic search but not strictly related
to SNS context, there exist various proposals and
solutions as illustrated by the survey of Mangold
(2007) and Wei et al. (2007). Ontology based
semantic search solutions (e.g. Bonino et al. (2004)
and Fang et al. (2005)) as well as ontology based
query expansion (e.g. Hoang & Tjoa (2006) and
Bhogal et al. (2007)) have enhanced techniques for
semantic search applications. In order to
implementing a solution and make improvements to
a search engine of an ISN, future research of our
work includes a detailed observation of more
ontology based query expansion approaches in the
literature to using the ‘Semiotic Web ontology’
method. Other approaches (e.g. Tvarozek et al.
(2008)) have trying to take advantage of the ‘faceted
browsing paradigm’ and employ an integration
solution between semantic search and visual
navigation in graphs using the idea of social
networks.
Previous work conducted by Reis et al. (2010b)
ICEIS 2010 - 12th International Conference on Enterprise Information Systems
60
have develop a study on the challenges related to
search in ISN; the authors propose recommendations
for a search engine better suited to this kind of
system. To the best of our knowledge were not
found in the literature so far investigations that have
specifically focused on semantic search in SNS
considering aspects of accessibility and inclusion.
We agree that the development of a search engine on
an ISN should include these new challenges.
We also argue that the approach developed in
this paper can improve and expand methodologically
and technologically Semantic Web techniques, such
as Web ontologies, illustrating immediate and
practical results for better SNS search engines. This
approach differs from others, since the search
solution outlined tries to derive the term meanings of
the search and also the term meanings of the content
produced by users in the system, as information
from the users profile together with their behaviour
and roles. The solutions presented in literature so far
do not deal directly with this approach. Future
experiments with real users can show whether our
approach can bring promising benefits revealing
search results more suitable to the context of social
and digital inclusion, and also to SNS in general.
6 CONCLUSIONS
Social network systems may provide inclusive
access to digital information for people, creating
situations where the users’ diversity is respected and
the access difficulties minimized. This is the purpose
of the Inclusive Social Networks (ISN). It is
important to provide information retrieving in a
more natural way from the user’s viewpoint, with
results that make sense to people. Therefore more
suitable mechanisms for search should take into
account the meanings created, shared and used by
people in the social network.
This paper presented new perspectives for search
in ISN which consider the inclusive social context. It
showed the outcomes of an analysis about how to
improve a search mechanism considering aspects
related to the digital and social inclusion. We could
verify with real users, that semantic aspects can
make a difference for the users to reach information,
and that the current syntactic search engines are not
enough to an ISN context. Inspired on the practical
context of ISN users, this paper outlined a novel
method based on Semantic Analysis Method from
Organisational Semiotics to build more
representative Web ontologies, as well as
requirements of a software tool that materializes
these ideas as a process. Besides, it showed
strategies for using the proposed ontology with the
search engine to provide more precise search results.
As further work, the goal of this research is to
improve (in the implementation sense) the ideas
drawn up for the search mechanism described in this
paper. For that, we aim to develop the semi-
automatic tool for building ‘Semiotic Web
ontology’, as an extension of the SONAR tool,
including the heuristics and rule transformations to
build the OWL ontology aided from OC.
Furthermore we intend to develop a pilot
implementation of this search engine based on the
‘Semiotic Web ontology’ in the ‘VilanaRede
system, using and improving the strategies
mentioned in this paper; also the work involves new
practical experiments with real users, utilizing this
novel search mechanism in order to evaluate and
validate the solution with empirical results.
ACKNOWLEDGEMENTS
This work is funded by Microsoft Research -
FAPESP Institute for IT Research (proc. n.
2007/54564-1) and by CNPq/CTI (680.041/2006-0).
The authors also thank colleagues from CTI, IC –
Unicamp, NIED, InterHAD and Casa Brasil for
insightful discussion.
REFERENCES
Baranauskas, M. C. C., 2007. e-Cidadania: Systems and
Methods for the Constitution of a Culture mediated by
Information and Communication Technology.
Proposal for the Microsoft Research-FAPESP
Institute.
Bhogal, J., Macfarlane, A., Smith, P., 2007. A review of
ontology based query expansion. Information
Processing & Management, Vol. 43, No. 4, pp. 866-
886.
Bonacin, R., Baranauskas, M. C. C., Liu, K., 2004. From
Ontology Charts to Class Diagrams: semantic
analysis aiding systems design. In: Proc. of the 6th
International Conference on Enterprise Information
Systems. Porto, Portugal.
Bonino, D., Corno, F., Farinetti, L., Bosca, A. 2004.
Ontology Driven Semantic Search. WSEAS
Transaction on Information Science and Application,
Issue 6, Vol. 1, pp. 1597-1605.
Boyd, D. M. & Ellison, N. B., 2008. Social Network Sites:
Definition, History, and Scholarship. Journal of
Computer-Mediated Communication. Number 13. pp.
210-230.
NEW PERSPECTIVES FOR SEARCH IN SOCIAL NETWORKS - A Challenge for Inclusion
61
Choudhari, A., Jain, M., Sinharoy, A., Zhang, M., 2008.
Semantic Search in Social Networks. <http://
www.cc.gatech.edu/projects/disl/courses/8803/2008/pr
oject/project_deliverables/group22/proposal.pdf>.
Accessed on January, 2010.
Fang, W., Zhang, L., Wang, Y., Dong, S., 2005. Toward a
Semantic Search Engine based on Ontologies. In Proc.
of the 4th International Conference on Machine
Learning and Cybernetics. Guangzhou, China.
Volume: 3, pp. 1913-1918.
Gürsel, A & Sen, S., 2009. Improving search in social
networks by agent based mining. In Proc. of the 21st
International joint conference on Artificial intelligence
table of contents. Pasadena, California, USA. pp.
2034-2039.
Haynes, J. & Perisic, I., 2009. Mapping Search Relevance
to Social Networks. In Proc. of 3rd Workshop on
Social Network Mining and Analysis. International
Conference on Knowledge Discovery and Data
Mining. Paris, France.
Hoang, H. H. & Tjoa, M., 2006. The State of the Art of
Ontology-based Query Systems: A Comparison of
Existing Approaches. In Proc. of IEEE International
Conference on Computing and Informatics (ICOCI).
Kuala Lumpur, Malaysia.
IBGE, 2008. Acesso à Internet e posse de telefone móvel
celular para uso pessoal 2008. Retrieved January,
2010 from <http://www.ibge.gov.br/home/estatistica/
populacao/acessoainternet2008>
IPM, 2007. Indicador Nacional de Analfabetismo
Funcional. Retrieved January, 2010 from
<http://www.smec.salvador.ba.gov.br/site/documentos/esp
aco-virtual/espaco-dados-
estatisticos/indicador%20de%20analfabetismo%20fun
cional%202007.pdf>.
Liu, K., 2000. Semiotics in information systems
engineering, Cambridge University Press.
Maedche, A. & Staab, S., 2001. An Ontology Learning for
the Semantic Web. IEEE Intelligent Systems. Volume
16, Issue 2. pp. 72 – 79.
Mangold, C., 2007. A survey and classification of
semantic search approaches, Int. J. Metadata,
Semantics and Ontology, Vol. 2, No. 1, pp.23–34
MEC, 2007. Retrieved November, 2009 from
<http://portal.mec.gov.br>.
Medelyan, O., 2006. Semantically Enhanced Automatic
Keyphrase Indexing. In Proc. of Women in Machine
Learning (WiML) Workshop co-located with the
Grace Hopper Celebration of Women in Computing.
San Diego, USA.
Neris, V. P. A., Almeida, L. D., Miranda, L. C., Hayashi,
E., Baranauskas, M. C. C., 2009. Towards a Socially-
constructed Meaning for Inclusive Social Network
Systems. In Proc. of International Conference on
Informatics and Semiotics in Organisations - ICISO.
Beijing, China. pp. 247-254.
Nielsen, 2009. Global Faces and Networked Places. The
Nielsen Company USA. Retrieved July, 2009 from
http://blog.nielsen.com/nielsenwire/wp-content/uplo
ads/2009/03/nielsen_globalfaces_mar09.pdf.
Peirce, C. S., 1931-1958. Collected Papers. Cambridge,
Mass: Harvard University Press.
Reis, J. C., Bonacin, R., Baranauskas, M. C. C., 2010a.
Analysing Search in Inclusive Social Network
Services. IC-10-02, www.ic.unicamp.br/~reltech.
Reis, J. C., Baranauskas, M. C. C., Bonacin, R., 2010b.
Busca em Sistemas Sócio-Culturais Online: Desafios e
Recomendações (Paper submitted to XXXVII
Seminário Integrado de Software e Hardware -
SEMISH 2010) (In Portuguese).
Ribeiro, L. C. & Vieira, R., 2008. OntoLP: Engenharia de
Ontologias em Língua Portuguesa. In Seminário
Integrado de Software e Hardware - SEMISH. Anais
do XXVIII Congresso da Sociedade Brasileira de
Computação. Porto Alegre: SBC, v. 1. p. 1-15. (In
Portuguese).
Santos, T. M., Bonacin, R., Baranauskas, M. C. C.,
Rodrigues, M. A., 2008. A Model Driven Architecture
Tool Based on Semantic Analysis Method. In Proc. of
10th International Conference on Enterprise
Information Systems. Barcelona, Spain. V. 2 pp. 305-
310.
Stamper, R. K., 2001. Organisational Semiotics:
Informatics without the Computer? In Information,
Organisation and Technology: Studies in
Organisational Semiotics, eds. K. Liu, R. Clarke, P. B.
Andersen and R. K. Stamper. Kluwer Academic
Publishers.
Tanasescu, V. & Streibel, O., 2007. Extreme Tagging:
Emergent Semantics through the Tagging of Tags. In
International Workshop on Emergent Semantics and
Ontology Evolution (ESOE2007). Busan, Korea.
Tvarožek, M., Barla, M., Frivolt, G., Tomša, M., Mária,
B., 2008. Improving Semantic Search via Integrated
Personalized Faceted and Visual Graph Navigation.
In SOFSEM 2008: Theory and Practice of Computer
Science. Lecture Notes in Computer Science -
Springer Berlin / Heidelberg. Vol. 4910, pp. 778-789.
Vieira, M. V., Fonseca, B. M., Damazio, R., Golgher, P.
B., Reis, D. C., Neto, B. R., 2007. Efficient Search
Ranking in Social Networks. In Proc. of 16th ACM
Conference on information and knowledge
management. Lisbon, Portugal. pp. 563-572.
Wei, W., Barnaghi, P. M., Bargiela, A., 2007. The
Anatomy and Design of a Semantic Search Engine.
Tech. rep., School of Computer Science, University of
Nottingham Malaysia Campus.
W3C, 2004. “OWL-Web Ontology Language”,
Recommendation 10 February 2004. Retrieved March,
2009 from http://www.w3.org/TR/owl-features.
ICEIS 2010 - 12th International Conference on Enterprise Information Systems
62