Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

GeoJournal (2008) 72:229–237

DOI 10.1007/s10708-008-9182-4

A trust and reputation model for filtering and classifying


knowledge about urban growth
Mohamed Bishr Æ Lefteris Mantelas

Published online: 29 July 2008


 Springer Science+Business Media B.V. 2008

Abstract In this paper we present a trust and Introduction


reputation model to classify and filter collaboratively
contributed geographic information. We hypothesize This paper proposes a trust and reputation model for
that users contribute information in a collaborative Collaboratively Contributed Geographic Information
system akin to Web 2.0 collaborative applications. (CCGI). The term CCGI is closely related to Volun-
We build on previous work where trust is proposed as teered Geographic Information (VGI), and it refers to
a proxy for information quality and propose a spatial individual observations by the users in the model
trust model to filter and extract high quality infor- presented in this paper.
mation about urban growth behaviors contributed by The central problem tackled in this paper is the
users. The motivating scenario involves residents of collection, filtering and classification of CCGI, using
recently urbanized areas taking into account their contributed information about urban growth as a
interactions with their surroundings. The main con- potential scenario. The term modeling refers to
tribution of this paper is a formal trust and reputation projecting a phenomenon or a process to a simplified
model that takes into account the spatial context of sub-space of the real world, thus resulting in a rigid,
users and their contributions. less dimensional analog of the phenomenon or the
process. Models are simplified versions of objects,
Keywords Social networks  Trust  conditions or processes (Ness and Low 2000). In
Reputation  Urban growth  Residential choice terms of efficiency and comprehensibility, modeling
involves choosing the smallest set of objects and the
smallest set of the simplest relations that reproduce
the patterns of change in a way that is accurate and
consistent with the evolution of the real phenomenon.
When presenting our trust and reputation model, it is
M. Bishr (&) tempting to try to capture all aspects of the problem
Institute for Geoinformatics, University of Muenster,
of determining CCGI quality in one model. Such a
48151 Muenster, Germany
e-mail: m.bishr@uni-muenster.de comprehensive model is unattainable at the outset, so
we adopt a layered approach and present the first
L. Mantelas layer in this paper. Our layered approach offers an
Regional Analysis Division, Institute of Applied and
initial model that can be expanded by inclusion of
Computational Mathematics, Foundation for Research and
Technology-Hellas, 71110 Heraclion Crete, Greece additional user behaviors, to incrementally increase
e-mail: mantelas@iacm.forth.gr the complexity of the model. For example, we

123
230 GeoJournal (2008) 72:229–237

exclude a temporal dimension from our current can be referred to as an information community. This
model, while acknowledging its importance in a information community possesses information that is
more comprehensive model. geospatial and temporal in nature. We are motivated
In Bishr and Kuhn (2007) we presented a vision of here by the notion of geospatial information commu-
CCGI, termed ‘geospatial information bottom-up’, nities, originally proposed by the Open Geospatial
that builds on the lessons learned from Web 2.0 (also Consortium (OGC) a decade ago. Information com-
sometimes called the Social web). Web 2.0 (Oreilly munities are defined here as ‘‘a community of
2005) is centered on the idea of harnessing the geodata producers and users who share a common
collective intelligence of information communities. set of feature definitions and ontology of real world
Web users are organized in virtual communities phenomena’’ (Bishr et al. 1999, p. 58).
centered on commonalities of interest. These infor- Having asserted that harnessing information com-
mation communities collaborate and produce large munities is the basis of our trust and reputation
amounts of information, including Geospatial Infor- model, we must point out that trust from a sociolog-
mation (GI). Bishr and Kuhn (2007) also identified ical point of view is a prerequisite for the existence of
many of challenges posed by collaborative community. Functioning societies rely strongly on
approaches to GI collection. Among those challenges trust between the individuals (Sztompka 1999; Usla-
is information quality. In large collaborative envi- ner 2002; Seligman 1997). On the one hand, in real
ronments, there are not yet quality measures to filter world communities citizens rely on subjective mea-
and classify the information contributed by individ- sures of trust. On the other hand, in online
uals, many of whom are not GI experts. The model communities objective measures of trust have been
we develop here rests upon a certain understanding of used as users rate each other with trust values
expertise. We assume that the contributors of CCGI pertaining to how much one user trusts another in a
are ordinary users with no specific training on GI. certain context since trust is a highly contextualized
Our approach to developing ways of evaluating the phenomenon (Zaihrayeu et al. 2005a, b; Golbeck
quality of the GI they contribute borrows methods 2005). Many definitions of trust exist, and we adopt a
that have proven effective on the web. Specifically, simple definition, viewing trust as a bet that an
we propose using trust and reputation as a proxy for individual makes about the future contingent actions
GI quality, while extending the concept with the idea of others (Sztompka 1999). Since our model depends
of spatio-temporal trust and reputation. on some form of social networks, four properties of
trust can be identified, namely, transitivity, compos-
Information communities and trust ability, personalization and asymmetry (Golbeck
2005).
Using trust and reputation to evaluate information Transitivity is about the propagation of trust
quality implies the existence of a community within through chains of people from one person to the
which trust and reputation are established. In our other, although trust is not mathematically perfectly
model, we subscribe to a certain notion of commu- transitive (Golbeck 2005). Composability is about
nity, that of an information community. Modeling is situations where different actors rate the same person
an inherent procedure in our lives since people use or item. From these ratings, we can then compose the
their own personalized empirical model of the world value and establish an absolute rating of that person
in order to make decisions. These models are built on or item. Personalization asserts that trust is relative,
people’s experiences and each single experience they and dependant on personal perspective. Given our
have tests the adequacy and consistency of their adopted definition of trust, a good outcome for one
models of the real world, expanding or altering their person could be a bad outcome for another. Asym-
knowledge base. In that sense the collective of these metry pertains to the fact that trust might not be equal
personalized views represents a very good picture of in both directions. That is, Person A might trust
reality, where each person’s view constitutes a small Person B independently of how Person B trusts
piece of a bigger picture that we term here the Person A. Our computational model exhibits all of
‘‘collective empirical model’’ of the world in which a these properties of trust through the formalism
certain group of people lives. This group of people presented.

123
GeoJournal (2008) 72:229–237 231

One important point to address is our use of the patterns of human behavior is a major challenge in
term ‘‘spatial’’ trust model. The question this raises is building efficient, more realistic urban growth models
why do we assert a relationship between the spatial (Cheng and Masser 2003). Existing Knowledge Base
location of an information contributor and trust? Extraction (KBE) methods that are based on the
Trust and reputation in our model occur over social analysis of aggregate data also fail to do so. They
networks. Studies have shown a direct relation cannot express the overall dynamic of a social system
between the evolution of social networks and geo- by mapping its partial dynamics. They lose the social,
graphic space (Metcalf and Paich 2005). Studies of temporal and decision-making heterogeneity that
how similarity between people breeds connection, exists in people’s behavior patterns. Furthermore,
also known as homophily (McPherson et al. 2001), even if we could reproduce the exact patterns that
showed that geographical proximity is a basic source occurred during the past, we can only apply them in a
of homophily. Zipf (1949) justifies this relationship model under the assumption that these patterns will
as a matter of effort, noting that it takes less effort to not change. We may capture the tendencies, but not
build and maintain relations when the subjects are in necessarily the potentials. While Berry (2004)
a close geographical proximity. Also some research focuses on the way households make decisions on
has used geographic distance as an indicator of an aggregate level, we need to investigate the internal
network density (Buskens 2002). Although technol- decision making process of individual households.
ogies have apparently loosened the effect of The problem then is to build a collaborative KBE
geography by lowering the effort involved (Kaufer system, which leverages the collective intelligence of
and Carley 1993), communities still show strong the information community to collect up-to-date
geographic patterns (Verbrugge 1983). Also, tech- dynamic information about urban growth in a contin-
nologies have allowed people to make homophilic uous, evolving manner. In such a system, the question
relations through other dimensions (Wellman et al. we need to answer is ‘‘How and why should we trust a
1996). certain individual piece of CCGI provided by a certain
Our spatial trust and reputation model combines user?’’ The provided information in our case is local
trust and the spatial dimension to filter CCGI, to knowledge from residents about recently urbanized
select the best candidate information to form a areas. This information can assist modelers in under-
‘‘collective empirical model’’ of the universe of standing the micro-level dynamics of urban growth. Our
discourse for a certain information community. In approach builds on a spatio-temporal trust model to filter
this paper we view the residents of urban areas as one through the potentially large flow of CCGI provided by
such information community. the users, to extract the highest quality information.

Problem description Problem approach

Understanding urban growth in terms of actors’ Cecchini and Rizzi (2001) suggest an urban analysis
behavior requires a collection of certain types of perspective to understand urban growth: ‘‘walk in the
data from residents living in areas of study. This streets and interact with the inhabitants, and discover
collected information reflects the residents’ view of that those who move across the city do so for
the area and the forces that are shaping urban growth different reasons, each person for more than one
through their residential choice. These information reason’’. We suggest the utility of walking directly to
entities constitute the CCGI in question for this paper. the source of the urban growth phenomenon: urban
As Agarwal et al. (2000) points out, models residents themselves. Let the decision-makers (resi-
incorporating higher levels of human-decision-mak- dents) model their own behavior and point out the
ing are more centrally located with respect to spatial important parameters and the way these parameters
and temporal scales. That is, they fall within the affect them, as well as explain the ‘‘reasons for’’ and
middle ground of those scales, probably due to the ‘‘the reasons not to’’ take particular actions. If we
lack of data availability at more extreme scales. extract this kind of knowledge from people, then we
Extracting information that captures micro-level can map the very specific reasons that lead them to

123
232 GeoJournal (2008) 72:229–237

particular behaviors that affect urban growth pro- example is based on an urbanization scenario of
cesses. There is an implicit assumption here that the areas that are showing rapid urban growth over
residents have a collective sense of the urban growth relatively short periods of time.
phenomenon, an assumption that is supported by the Urban modelers study an area that has been rapidly
notion of information communities (Bishr et al. urbanized within the last decade in order to under-
1999) as discussed earlier. The methodologies of stand the micro-level dynamics behind urban growth.
earlier approaches failed to capture micro-level The modelers need to find a theoretical framework
details about people’s behaviors. A number of the that fits the actual facts and sufficient data to
previously presented approaches incorporate inter- implement and calibrate it. Instead, they may view
view and questionnaire features. Luo and Sen (2004) the community of residents within this urban growth
and Otoo et al. (2006) made field surveys using area as an information community in which each
questionnaires, formal and informal interviews, and individual has their own model of the world and their
focus group discussions to explore and understand the experience with urban growth. The collective view of
nature of the spatial growth processes. These this community forms a collective empirical model of
approaches though, took into account certain groups their world.
of businessmen, peasants and local officials, and Our scenario would include a system for collecting
included a limited number of interviews and discus- several elements of users’ experiences of urban
sions processes while aiming more at the economic growth as CCGI. Users would be provided with
level of analysis. web interfaces and handheld devices with which they
We propose a wider scale of application that relies could log information pertaining to the criteria
on the foundations of Web 2.0 and overcomes the mentioned earlier, such as:
limitation of the Otto’s and Luo’s approaches. By
• marks for locations of urban sprawl around urban
building web applications utilizing CCGI gathered
boundaries
from the residents of growing urban areas, we can gain
• buildings that have recently been built
insights about the micro-level patterns of human
• parameters of recently built buildings (floors,
behaviors and use this information to deliver better
heights, etc.)
services. Initially, we need to focus on the institutional
• estimated income averaging for certain locations
framework for the information required and define a set
in the urban areas
of specifications concerning the type and form of the
• a variety of neighborhood characteristics (traffic
CCGI we collect. Barredo et al. (2003) identifies five
jams, public transport hubs, etc.)
fundamental aspects of urban growth that should be
included in the institutional framework: Residents could collect and mark this information
and possibly any other parameters deemed necessary
• environmental characteristics,
by the urban growth modelers using the map-assisted
• local-scale neighborhood characteristics,
web forms or handheld devices. The forms and
• spatial characteristics,
devices would require the residents to provide other
• urban and regional planning policies and
specific information to help the system calculate trust
• factors related to individual preferences, level of
ratings for filtering later. This information would
economic development, socio-economic and
include the following elements:
political systems.
• the location of the residents at the time where they
CCGI about urban growth should touch on these
noticed the information they contributed as CCGI
elements to ensure coverage of the phenomenon
• if a particular CCGI was previously reported by
within the reported information by the observers.
some other resident, then the resident attempting
to report it again would have to give the previous
Motivating scenario reporting a rating. The rating pertains to the
quality of the previous reporting as an accurate
In this section we present a scenario that demon- representation of reality. Giving a rating would
strates the model presented in this paper. The not preclude the user from re-reporting the

123
GeoJournal (2008) 72:229–237 233

information again on their own if the user has but argue that if some trust rated geospatial informa-
better information tion is useful and relevant to larger group of users it
• the system could count how many times the same can be assumed to have satisfactory quality in a more
CCGI entity was reported by different users. objective sense (Bishr and Kuhn 2007).
In Bishr (2007) we introduced a spatio-temporal
With the information provided to this system, our
trust model for social networks. The model presented
trust and reputation model could be applied to
here is different from its predecessor in that it takes into
compute trust ratings for the contributed CCGI from
account ratings of information contributions and not
all the users of the system. The urban growth
just the number of times an information entity is
modelers could then study the micro-level factors
reported. It also uses a log function for distance rather
behind the urban growth process using the extracted
than raw distance. These two modifications improve
information. In the following section we illustrate
the ratings of the model. First, by incorporating ratings
how our trust and reputation model functions to
of the information entities made by its consumers, the
achieve this required rating and filtration of CCGI.
model has an explicit account of trust ratings that was
not available in the earlier approach. This explicit
account is important since we assume that the objective
Spatio-temporal trust and reputation model quality of an information entity is directly proportional
to the trustworthiness of the information entity, as
Extensive research has been done on trust in web based judged by information consumers. In the earlier model
social networks (WBSN) for a variety of applications, trust was not explicit. Second, the use of raw distance in
including email filtering and web based recommender the earlier model was problematic because large
systems (Golbeck 2005). In this research, trust calcula- distances (e.g. a contributor living in a suburb at the
tions have proved superior in providing a measure of the outskirts of a city) resulted in trust values that were
quality of movie reviews and emails, when compared to disproportionally small compared to people living
both the average rating and ratings generated by shorter distances from the center. To smooth the trust
traditional collaborative filtering algorithms. Some function we resorted to a simple log function that
web researchers are also interested in using social yielded more uniform results across sufficiently large
notions of trust to identify how answers provided on the distance variations.
web can be trusted. The Inference Web (McGuinness It should be noted that our model in this stage does
et al. 2004) aims to take vague query answers and make not account for the temporal dimension of some
them more transparent by providing explanations. phenomena. Some features change continuously,
Zaihrayeu et al. (2005a, b) provide an extension of the such as traffic information or pollution, and so some
inference web that is termed IWTrust. IWTrust uses locations would be associated with temporal variables
quantified measures of trust in answers returned by in the model. As noted earlier, we adopt a layered
search queries on the web. It utilizes trust values in approach to expanding model complexity and do not
social networks among users, and trust values between include temporal variables at this stage. We acknowl-
the users and the information to compute these trust edge the importance of integrating such variables and
ratings. leave that for future work.
Trust in WBSNs measures the value of informa- When a resident marks a certain piece of CCGI the
tion produced by a group of users to others users who process result can be viewed as a bipartite graph
consume this information (Ziegler and Lausen 2004; (Fig. 1). The resulting bipartite graph of people
Golbeck 2005; Zaihrayeu et al. 2005a, b). Users who (actors) and contributions is a two mood non-dyadic
are trusted by others (aka. trustees) are the ones who, affiliation network (Fig. 1). Actors are possibly
from the perspective of the trustors, provide more affiliated to many contributions and contributions
useful information. Generally in CCGI environments are possibly contributed by many actors. Hence, our
there is a lack of traditional quality criteria such as affiliation network consists of a set of actors and a set
lineage, accuracy, consistency and completeness. We of contributed information entities. We observe two
propose using trust as a proxy for GI quality. We sets of nodes representing the two moods of the
view quality as a subjective measure to some extent, network shown below.

123
234 GeoJournal (2008) 72:229–237

N: is the set of all actors (people) who are The first step is to calculate a trust rating based on
contributing information
 to the network.
 Those actors the number of people contributing CCGI and
can be viewed N ¼ n1 ; n2 ; . . .; ng in terms of the weighted by two factors:
subset of CCGI to which they contributed. M: Is the
• a rating given by actors N to CCGI entities M, this
set of all CCGI that are contributed by actors n,
rating is denoted rn,m and is on a scale of 1–10.
similar to the members of the set N, CCGI can also be
• the distance between an actor and the CCGI they
viewed in terms of the subset of M ¼ fm1 ; m2 ; . . .;
contributed, such a distance measure is an indi-
mh g actors who contributed these CCGI. In Fig. 1 a
cation of the network density (as defined in
link exists between actors/person n2 for example and
Wasserman and Faust 1994). Network density has
CCGIm3 when the person has marked, rated, and/or
an inverse relation with distance, while (Nohria
reported the information entity.
and Eccles 1992; Buskens 2002) suggested a
Also between actors N we take into account a
positive relation between network density and
social network representing the relationships between
trust. This positive relation is an artifact of the
people in a study area. Such relations include but not
speed of transmission of information in networks
limited to, kinship, neighborhood and friendship. Our
and strength of relations between actors, which in
model has a hybrid network structure that represents
turn fosters trust.
an affiliation network between people and CCGI
entities, as well as a social network between people The proposed first component of the trust rating is
involved in the system. The problem is now to based on the nodal degree d(mh) (Wasserman and
calculate the trustworthiness of a given information Faust 1994). The first component of the model is then
entity such as m3 based on the reputation of the users denoted 0 d(mh) such that:
who contributed this entity.
Xk rðn;mÞi
0
dðmh Þ ¼ where ci [ 1 ð1Þ
i¼1 logðci Þ
where 0 d(mh) is an adjusted nodal degree for a certain
CCGI such as m3 taking into account the two factors
mentioned above and r(n,m)i is the ith rating from n to
m and Ci is the distance between n and m at the time
that n reports the CCGI entity m. we opted for a log10
function to smoothen the sharp variations in the
equation implied by using raw distance.
The second two equations are straightforward.
Equation 2 represents a trust calculation at the social
network level between the actors.

P
N
ðtni ng Þðkni ng Þ
ni 2adjðng Þ;i¼1
tng ¼ ð2Þ
N
where tng is the trust rating of an actor n based on the
trust ratings given to him/her by the actors with
whom he/she is connected in the network, tni ;ng is the
rating from actor i to actor g and kni ;ng is a rating that
represents the strength of social ties between the
Fig. 1 An affiliation network of agents/users (white nodes) actors. This rating is personal and should be deter-
and the CCGI they contribute (black nodes) with social mined based on user opinions and analysis of the
network links weighted by distance and user ratings. Lines
nature of relations such as kinship or friendship for
connecting N nodes are a separate agent-agent social network,
making this model a hybrid affiliation-one mode social example. Based on Eq.2 the final trust rating of the
network information entity m is given by Eq. 3

123
GeoJournal (2008) 72:229–237 235

Conclusions
Xk
tng rðn;mÞi
tmh ¼ where ci [ 1 ð3Þ
i¼1;g¼1
logðci Þ Web 2.0 encompasses the key ideas of harnessing the
collective intelligence of web communities, enabling
Equation 3 introduces the final trust rating of a interaction and collaboration between web users
certain CCGI contributed by a resident of an urban organized in communities centered on common goals
growth area. The metric presented is sensitive to and objectives. Users are no longer only content
geographic distance, and the nature of social consumers, but are also content producers. As early in
relations, trust ratings and distance. By filtering the web’s history as 1996 studies of the quality of
information using this metric we can extract user-provided information proved that large amounts
relevant CCGI related to the area of study and of information were simply unreliable, outdated or
that are of adequate quality, since the model is plainly wrong (Janes and Rosenfeld 1996). The case
premised upon the notion that the quality of the becomes more evident in Web 2.0 applications, since
information is strongly determined by its spatial/ contributors are a heterogeneous group of people with
social context. varying experiences and backgrounds, resulting in
One question that remains unanswered is how the large amounts of information that is lacking tradi-
model will behave in instances where users submit tional quality criteria. Traditional models of authority
CCGI about a location where they made a direct (libraries, publishers, editors, media outlets) that
observation, but they are not presently located. For controlled the flow of information are being chal-
example, a user might report information about the lenged. New models of authority need to emerge to
street where s/he lives while s/he is on a vacation control the flow of information in Web 2.0.
elsewhere. Our model assumes the location of the Into this context, our paper contributes a commu-
observer at the moment he is making the observation nity-based collaborative model of authority, one
and does not deal with such exceptional situations. where the producers make information, and their
However, our model could be easily adjusted by peers and other information consumers judge the
changing the definition of location, without altering information quality. A valid metaphor here is the
the formal model itself. This can be done by scientific peer review process, albeit on a much larger
assuming the location used in the model to be the community scale and in a Web 2.0 collaborative
observer’s usual location, where s/he generally has manner. We propose using community based infor-
knowledge, instead of the location while actually mation authority to assure quality of information,
making the observation. mainly based on trust and reputation. Using trust and
The Eqs. 1–3 represent the core of our formal reputation as a measure of information relevance and
model. They depend on the existence of a network quality is well-documented in web research, and our
model as in Fig. 1 to give trust ratings for CCGI approach proposes a novel method to include spatial
tokens. By taking into account the location of and temporal parameters in making trust judgments
observers when they make observations, ratings by about geoinformation. Our trust and reputation model
the system users, and bi-partite graph properties can be applied to filter and extract information about
(count of contributions) we can compute trust ratings the residential choice process, to better understanding
for the CCGI observations. In a system where such a its effect on urban growth.
model is implemented, the trust-rated tokens can be One important element of our model must be
filtered out and served to information consumers as assessed is its resilience to improper tampering or
reliable information. fraud. Some users might abuse the system and
Traditional models of information authority are provide false or irrelevant ratings either on purpose
generally lacking in VGI environments and thus, or inadvertently. Such a problem is inherent in all
alternative models are required. Our model builds a trust and reputation systems. One answer is that the
collaborative authority system that corresponds with average of the volume of contributions will adjust
the collaborative nature of VGI. Further research on itself to proper values overcoming improper usage of
other areas of VGI where this model could be of the system, and this view is supported by existing
benefit is an ongoing effort. research (Surowiecki 2004). However, further

123
236 GeoJournal (2008) 72:229–237

research into this issue is essential to ensure model questions of privacy, especially how these are to be
resilience to potential abuse by some users. handled technically, not just legally or institutionally.
Goodchild (2007) proposes the term VGI to Our model brings up at least one issue with respect to
describe a collection of applications and web phe- privacy in the context of VGI. Our approach for
nomena that involves large-scale collaboration of assessing the quality and integrity of contributed
heterogeneous users. Compared to geogspatial data information rest upon having some information about
developed and managed in more conventional ways, the contributors, in this case their residential location.
such as through spatial data infrastructures (SDIs), This example illustrates that issues of privacy asso-
the phenomenon of VGI is unique in several ways. It ciated with VGI might emerge not only from the
has emerged from the bottom up and does not tend to contributed information itself, but also from the
rely upon top-down approaches to ensuring informa- models and methods that are developed and
tion quality or data sharing. Many traditional models employed for working with this unique new source
of information authority that have been used with of geographic information.
more conventional forms of spatial information are
not present or useful with VGI. Identifying these and Acknowledgement The authors would like to thank the
anonymous reviewers who have provided constructive
other gaps between VGI and traditional GIS is an feedback. We also would like to thank Sarah Elwood for her
important step in the emerging VGI research agenda. efforts in putting this publication in shape. This work is
Identifying these gaps begins to highlight key supported by COMPASS project (COastal Marine Perception
research needs and directions, particularly with Application for Scientific Scholarship), financed by
e-research:e-Information program of JISC (Joint Information
respect to the hardware, software, methodological,
Systems Committee) in the United Kingdom.
institutional framework, and data handling issues that
will need to be addressed if we are to make use of
volunteered information. References
The scenarios we have developed in this paper
contribute to these efforts to identify key research Agarwal, Ch., Green, G. M., Grove, J. M., Evans, T. P., &
directions in VGI, and respond specifically to the Schweik, Ch. M. (2000). A review and assessment of
land-use change models: Dynamics of space, time, and
need for techniques that are suitable for working with human choice. In 4th International Conference on Inte-
these data, in light of their unique properties grating GIS and Environmental Modeling, Problems,
compared to other forms of spatial information. Prospects and Research Needs. Banff, Alberta, Canada,
Based on the expectation that VGI will be tremen- 2–8 September 2000.
Barredo, J., Kasanko, M., McCormick, N., & Lavalle, C.
dously diverse (because of the diverse nature of its (2003). Modeling dynamic spatial processes: Simulation
contributors) and that the quality of contributed of urban future scenarios through cellular automata.
information will vary widely (because of the range Landscape and Urban Planning, 64, 145–160.
of skills and experience of the contributors), we offer Berry, B. (2004). A theoretical model for measuring the
influence of accessibility in residential choice behaviour.
a model for information filtering and classification. In 44th European Congress of the European Regional
Our model might be applied to extract the contributed Science Association, Porto, Portugal.
information expected to be most reliable, from a more Bishr, M. (2007). Weaving space into the web of trust: An
varied collection of information. Thinking beyond asymmetric spatial trust model for social networks.
CSSW. Leipzig, Germany: Bonner Köllen Verlag.
our specific technique described here, we would Bishr, M., & Kuhn, W. (2007). Geospatial information bottom-
argue that the emerging research agenda must up: A matter of trust and semantics. AGILE. Aalborg,
develop many such techniques for working with Denmark: Springer.
VGI. Specifically, the phenomenon of VGI centers Bishr, Y. A., & Pundt, H., et al. (1999). Probing the concept of
information communities-a first step toward semantic
upon generating and organizing information based interoperability. In M. F. Goodchild (Ed.), Interoperating
upon its geographical properties, and further research geographic information systems (pp. 55–70). Springer.
must consider the implications of this for information Buskens, V. W. (2002). Social networks and trust. Dordrecht:
archival and retrieval systems, information discovery, Springer.
Cecchini, A., & Rizzi, P. (2001). Is urban gaming simulation
location-based services, and existing techniques for useful? Simulation Gaming, 32(4), 507.
assessing information quality. Finally, in a broader Cheng J., & Masser, I. (2003). Understanding urban growth
societal sense, VGI research much also examine system: Theories and methods. In 8th International

123
GeoJournal (2008) 72:229–237 237

Conference on Computers in Urban Planning and Urban Otoo, E. A., Whyatt, D. J., & Ite, U. E. (2006). Quantifying
Management, Sendai City, Japan. urban growth in Accra metropolitan area (Ama), Ghana
Golbeck, J. A. (2005). Computing and applying trust in web- and Exploring Causal Mechanisms, England Promoting
based social networks. Department of Computing, PhD Land Administration and Good Governance 5th FIG
thesis, University of Maryland. Regional Conference Accra, Ghana.
Goodchild, M. F. (2007). Citizens as sensors: The world of Seligman, A. B. (1997). The problem of trust. Princeton Uni-
volunteered geography. GeoJournal, 69(4), 211–221. versity Press.
Huapu, L. (2002). Review of the urban growth over the past Surowiecki, J. (2004). The wisdom of crowds: Why the many
twenty years and prospects for the next 2 or 3 decades. are smarter than the few and how collective wisdom
Institute of Transportation Engineering, Tsinghua shapes business, economies, societies, and nations. New
University. York: Doubleday.
Janes, J. W., & Rosenfeld, L. B. (1996). Networked informa- Sztompka, P. (1999). Trust: A sociological theory. Cambridge:
tion retrieval and organization: Issues and questions. Cambridge University Press.
Journal of the American Society for Information Science, Uslaner, E. M. (2002). The moral foundations of trust. Cam-
47, 711–715. bridge: Cambridge University Press.
Kaufer, D., & Carley, K. (1993). Communication at a distance: Verbrugge, L. (1983). A research note on adult friendship
The effect of print on socio-cultural organization and contact: A dyadic perspective. Social Forces, 62, 78–83.
change. Hillsdale, NJ: Lawrence Erlbaum. Wasserman, S., & Faust, K. (1994). Social network analysis:
Luo, X., & Sen, J. (2004). Cross-border urban growth: The case Methods and applications. Cambridge: Cambridge Uni-
of Jiangyin economic development zone in Jingjiang. In versity Press.
15th Biennial Conference on the Asian Studies Associa- Wellman, B., Salaff, J., Dimitrova, D., Garton, L., Gulia, M., &
tion of Australia, Canberra. Haythronwaite, C. (1996). Computer networks as social
McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds networks: Collaborative work, telework and virtual com-
of a feather: Homophily in social networks. Annual munity. Annual Review of Sociology, 22, 213–238.
Review of Sociology, 27, 415–444. Zaihrayeu, I., da Silva, P. P., & McGuinness, D. L. (2005b).
McGuinness, D. L., & da Silva, P. P., et al. (2004). IWBase: IWTrust: Improving user trust in answers from the web.
Provenance metadata infrastructure for explaining and Springer.
trusting answers from the web. Springer. Zaihrayeu, I., & da Silva, P. P., et al. (2005a). IWTrust:
Metcalf, S., & Paich, M. (2005). Spatial dynamics of social Improving user trust in answers from the web, Springer.
network. Evolution, 51, 61801. Ziegler, C. N., & Lausen, G. (2004). Spreading activation
Ness, G. D., & Low, M. M. (2000). Five cities: Modelling models for trust propagation. The IEEE International
Asian urban population-environment (pp. 43–67). Oxford Conference on e-Technology, e-Commerce, and e-Ser-
University Press: Dynamics. vice. Taipei, Taiwan.
Nohria, N., & Eccles, R. G. (1992). Networks and organiza- Zipf, G. (1949). Human behavior and the principle of least
tions: structure, form, and action. Boston, MA: Harvard effort. Menlo Park, CA: Addison Wesley.
Business School Press.
Oreilly. (2005). http://www.oreillynet.com/pub/a/oreilly/tim/
news/2005/09/30/what-is-web-20.html

123

You might also like