Evolution and Dynamics of Information Cities: Models, Taxonomy and Architecture

Evolution and Dynamics of Information Cities: Models,
Taxonomy and Architecture
J. Sairamesh
IAC, IBM T. J.
Watson Research
Hawthorne, NY, 10532
1-914-784-7372
jramesh@us.ibm.com
P. Kavassalis
ICS-FORTH
Heraklion
Greece
petros@itc.mit.edu


S. Haridi
SICS and KTH
Kista
Stockholm
seif@sics.se



ABSTRACT
Similar to physical cities, which emerged over many
years in the physical world, we envision many kinds of
large-scale “information cities” (iCities) to emerge over
the Internet in the coming years. Some of these iCities
will mimic small and large “physical” cities. They will
provide local and global information and various value-
added services to inhabitants. Just like physical cities,
these information cities will house “infohabitants”, who
will participate in activities such as searching,
transacting, socializing, and collaborating. In this
paper, we present an overview of the information cities
project, and then we present a taxonomy of information
cities and some dynamic models. We also describe
design requirements and an open architecture for
building an information city based on Webservices.
Categories and Subject Descriptors
D.3.3 [Digital Cities, Agent Cities]: Models, economics, design
and architecture.
General Terms
Algorithms, Design, Architecture, Economics, Experimentation
Keywords
Digital Cities, Information Cities, Marketplaces, Simulation, and
Taxonomy of Cities.
1. INTRODUCTION

We are currently witnessing a trend in the formation and
evolution of large-scale information sites providing services to
diverse participants over the Internet. Some of these information
sites are evolving from being traditional commerce sites and web-
portals to offering more complex, social and collaborative
environments to participants. We call these complex aggregations
and agglomerations over the Internet as “information cities”.
Some of these information cities boast populations of tens of
millions of members or “infohabitants”, larger than most physical
cities.
Driven by profit and non-profit objectives, many kinds of
information cities (e.g. large web-sites, e-Marketplaces, Web-
portals, community portals, digital cities, social networks and
others) have emerged, and are still emerging. Unlike physical
cities, the boundaries of these information cities are not restricted
by physical space or proximity, instead on the costs of
participation, familiarity of navigating the city, search and
transaction overheads, computational resources (e.g. disk size and
processing power) and other frictional boundaries, which will
limit the expansion of these information cities. We envision a mix
of humans and software agents (representing humans) to inhabit
these information cities.
Recently, new types of information cities have begun to emerge.
These are “local information cities”, owned by the local
governments and municipalities of physical cities. These cities
offer information and commerce services to inhabitants of their
respective physical cities, and they offer access to local
government services, city departments such as law enforcement,
police, department of transportation, birth-certificate departments,
and so on. An example of such a city is Virtual New York
(NYC.gov [17]), which is managed by the mayor’s office. This
information city provides direct access to 56 local municipal
departments. Since its launch, it has attracted thousands (if not
millions) of inhabitants of New York City and others to benefit
from these services. In this paper, we present a taxonomy of
information cities, and some behavioral models of information
cities. In section 5 we describe an open architecture of an
information city.
2. Goals and Approach

Our goals in the information city (iCities) project [22,23] are to
model, design and prototype information cities. Some of the
underlying questions we are seeking to understand are the
following: What are the analogies between information cities and
physical cities? What are the interactions between agents and
humans in information cities? What are the factors causing the
formation and stability of information cities? Will users switch
freely from one city to another over the Internet? What are the
“equivalent” costs involved for inhabitants of an information city
(similar to distance in physical cities)? What are the key design
principles for large-scale information cities? Models for
information economies were done in [4] [25]. By using similar
models and simulating such environments, we plan to demonstrate
the emergence of various information macro-structures, i.e.
agglomerations and other regularities, and socio-economic
networks.

2.1 Approach

Our research work is done in three-phases, as illustrated in Figure
1 below.

Figure 1: Approach

Our first phase involves modeling the behavior of information
cities, behavioral patterns of infohabitants, and competition
among cities. Parameters such as user-preferences, costs of
participation, costs of advertising, competitive and behavioral
rules are taken into account. The second phase involves
simulating the behavior of the information cities using simulation
tools (Mozart [8], [24]). We employ a multi-agent simulation
environment, where agents mimic the behavior of humans, by
executing simple behavioral rules [8]. The third phase is a design
phase to prototype simple agents with behavioral rules, and
information cities with real humans and software agents as
infohabitants.
3. Classification and Taxonomy

In the coming years, we envision many forms of information
structures and community networks of various sizes, degrees and
scales emerging over the Internet. We call these aggregations as
information cities. These “information cities” can be classified
based on the services they offer in a multi-dimensional space as
shown in the Figure 2. Large scale “information cities” such as
Yahoo [19] and AOL [20] provide a combination of many
services in order to attract consumers. In Figure 2, along each of
the axes, we represent a range of services, which vary in
complexity. The simple services are close the origin, and farther
away from the origin are the richer and more complex services.
For example, Yahoo mainly offers content services and
community services such as e-mail, discussion groups and forums,
and yellow-pages services, and recently some commerce services.

Classification of Information
City Services
Content services
Commerce and Business
Community and Social services
z
y
x
Static
content
Dynamic
Content
Advanced Navigation/Search and knowledge
bases
email
forums
collaboration
Increasing
complexity
Increasing
complexity

Figure 2: Classification

Using the above coarse classification, and details of each axis in
the classification, we study the various types of information cities
that current inhabit the Internet. A large body of literature exists
on Digital Cities, and is gaining momentum among many
disciplines such as urban planning, computer science, architecture
and others (Toru et. al [12], S. Graham[10], [11] and W.
Mitchel[13], [14] and others).
In the Webster dictionary, a city is defined as a “municipal
corporation whose powers are confined to a fixed area and subject
to the authority of the state”. This definition, though simple
enough, demonstrates that physical boundaries are the key to the
physical cities. We define an “Information City” as the following:
a large scale internet-based portal that offers a range of services
such as searching and matching services, interactive social
environments, community services, digital library services and
electronic commerce to its infohabitants (inhabitants). A physical
city embodies social norms and provides municipal services for
social stability. Will information cities have similar
representations? Some information cities are emerging to behave
like physical cities, and some are truly global, and will demand a
new way of governing activities.
3.1 Taxonomy

We classified information cities in the previous subsection based
on the services they offer to the “infohabitants”. The
classification, though coarse, provides a framework for comparing
the range of services being offered by various kinds of
information cities. The taxonomy is as follows for 3 kinds of
information cities.
Local cities: These are mainly clones of physical cities, providing
information about an existing physical city to the inhabitants.
These cities are owned by local governments, which provide
services for commerce with the government (e.g. such as paying
taxes, payment for city parking, accessing legal documents), and
provide other information about the rules and regulations of the
city, and access to public administration offices. Regional Cities:
a collection of digital cities represented through a single web site.
Examples include AOL Digital Cities[20], CitySearch Digital

Cities, Netherlands Digital Cities[16], and so on. These regional
and national cities provide a common set of services to inhabitants
interested in finding regional and local information such as
weather, news, entertainment, information about local and
regional events, cultural events, social places, and other attractive
services.
Commercial Cities: These are established e-commerce sites such
as eMarketplaces or Web-portals that provide a marketplace with
industry news, and discussion forums for traders, and a wide
range of e-commerce services such as auctions, catalogs, search,
RFQs (request for quotes), reverse auctions, order management
and payment services. Industry verticals: These are large-scale
web-based commercial portals that provide a common set of
commerce and business content services for buyers and sellers in
many industry verticals. Examples include VerticalNet[21].
Social community cities: These are web sites for social and
community based interactions. They offer a range of information
services and social interaction services such as email, chat rooms,
discussion forums and special interest lists, private discussion
rooms and other. Yahoo [19] and AOL [20] are examples of such
social cities. Local community web sites: These are these are
public web sites, owned by the city or state that provides all the
necessary information and community services to a specific
community.
4. Economic and Behavioral Models

In this section, we present a model of a large information city,
where the number of inhabitants: software agents and human is in
the millions. The model is based on a transaction model of
purchasing physical and information goods from marketplaces,
which provide an electronic means to access information and
place orders. The main reason behind the modeling effort is to
understand via agent-based simulations the behavioral patterns of
users, and survivability of information cities.
Consider that there are 1 to N cities that offer commerce and
information services. Assume for now that a marketplace (one or
more marketplaces can exist in an information city) for commerce
exists in each city. The marketplace is mainly a place in the
information city for trading goods and services. Each marketplace
supports one or more categories of product information (1, C),
and each category contains one or more products. The products
could be physical goods or information goods/services. Each
product has one or more attributes that describe the product.
Examples of attributes are: price, quantity, quality, delivery time,
discount terms and conditions based on how much has been
purchased. Each information city has a cost for maintaining the
information, and providing transaction services. This cost is
based on disk space, computational power, human resources,
agent resources, electrical utilities, physical storage space and so
on.
Users in this model can roam from information city to information
city (but are limited by participation costs). Users have
preferences on products and information they wish to purchase.
The preferences are based on the attributes of the products such as
price, quantity, delivery and others. The user utility functions
capture the constraints and relationships on the products. There is
cost for users to switch from one city to another, and a cost for
performing searches and transactions within a city. Users select
information cities based on what is being offered, the costs of
participation, the number of products supported, and information
on their preferred partners (e.g. suppliers, brand-name
manufacturers and others). Similarly, suppliers who offer
products also have preferences on products and costs of
participation.
In the model, a user will select an information city based on many
factors such as: products that are supported by the information
city, suppliers who are participating in this city, costs of
participation, and similarities of preferences of other users. The
assumption here is that users wish to find information cities that
have users who have similar preferences. Matching algorithms
are run by the information cities to match users with similar
preferences. The algorithms cluster users with similar preferences
based on attributes and constraints. Similarly, users (user-agents)
run search algorithms to find the best possible cities to inhabit. In
our model, a simple global yellow pages service provides
information on each of the cities. The cities advertise their prices,
products offered and participating suppliers at these yellow pages.
Simulations were done based on the above model. The simulation
environment is based on software agents that are programmed to
behave as users, suppliers and information cities. Each agent
executes specific rules of participation, selection and consumption
of goods and information. The simulations revealed the formation
of information cities, where groups of users with similar
preferences converged at information cities that matched their
preferences and criteria. The city formation1 was sensitive to the
initial conditions and preferences. In a majority of the simulation
configurations, the users converged to a few information cities as
long as the costs of participation, subscription and transaction
were comparable to the wealth of the users.
A sample example of the economic model is illustrated in the
Figure 4. User preferences (utility function) in the attribute space
are captured in figure (4a). Figure (4b) and (4c) illustrate the user
preferences matched with the offerings of iCities. The iCity
offerings are mainly the aggregate offerings of participating
suppliers in the same attribute space. In figure (4b), the matching
between user preferences and iCity is fairly high compared to the
matching in figure (4c). The simulation is done with a large
number of users and cities to study the evolution and dynamics.

1 Detailed results from the simulation are presented in an
upcoming paper.

Figure 3: User preferences, utility and matching
5. Architecture

In this section, we first present some of the core design issues for
building an information city that can support a variety of
infohabitants such as: consumers, businesses, administrators,
regulators, municipalities, special interest groups and others. We
propose an architecture based on Webservices, which enables an
open environment for searching, advertising, matching, finding
and invoking a range of capabilities.
5.1 Design Criteria

There are many design criteria to consider when creating,
developing and deploying an information city.
User preferences and customizability of content: content must be
customizable to different types of infohabitants based on
preferences. Content comes in various flavors such as product
information and offerings, real-time news and content, weather
reports, content from archives, digital libraries (representing
physical libraries) and other pertinent information.
Matchmaking services: A crucial design requirement is matching
of profiles and capabilities. Businesses looking for consumers can
be matched based on interest profile of the consumers and
offerings of the businesses.
Social interaction and collaboration: As a core design
requirement, the basic functionality such as chat, discussion
groups, public and private interest lists and other are crucial for
establishing social networks in the information city.
Commerce services: These include basic services such as
searching for businesses, catalog for businesses, catalog based
buying, auctions for trading and simple Request For Quotes
mechanisms. In addition, billing and payment are required for
fulfillment. The city can also house banks, payment centers,
insurance providers and others that provide day-to-day services
for infohabitants.
Standards for open cities: As large-scale information cities begin
to evolve, standards such as Webservices, integration protocols
such as SOAP (and SLAs2), middleware components, agent
communication languages will need to be incorporated in the
design to enable open interactions with humans and agents.
5.2 Architecture Components

5.2.1 Server Architecture

The architecture, shown in Figure 3, illustrates the various
components in the end-to-end connectivity and commerce
between various types of infohabitants.

Figure 4: iCities open architecture based on webservices


Each infohabitant in the design is represented by a profile in the
information city. A part of the profile is made public by the
registered infohabitant. Similarly, a business will have a
searchable business profile. The public information of the profile
is available in the yellow pages and accessible by search and
navigation engines. Users and business can advertise their
capabilities and let others search, find and invoke their services.
5.2.2 Middleware Runtime

The middleware layer consists of many sub-components (all of
which are not shown in the figure). The middleware handles the
various interactions between infohabitants and services offered
within the information city. The services offered are the
following: advertising, creating profiles, finding services, finding
supplier capabilities, messaging, transactions, directory services
(through UDDI3 protocols), and personalization. With the
emerging standards on Webservices, the creation of universal
directories, and communication protocols provide the necessary
building blocks for open architectures for public and private
information cities.

2 Service Level Agreements
3 UDDI stands for Universal Description Discovery and
Integration

6. Conclusion

In this paper, we presented a taxonomy of information cities based
on their structure and services offered to the various members
(infohabitants). We classify the information cities based on the
service offerings to the infohabitants. The classification and
taxonomy provided the necessary understanding of the design
criteria required for building medium to large scale, complex,
information cities. We presented a novel architecture of an
information city server that will support a wide variety of
infohabitants, and provide the necessary services for content
searching, matching making, commerce transactions and
community services.
We proposed a collection of core components of an information
city for fostering socio-economic interactions: directory services,
advertising and profiling of infohabitants, messaging, access
control, navigational structures, navigational user interfaces and
connectivity. We envision information cities to evolve in the
coming years, and their success depends on open architectures to
enable a variety of socio-economic interactions and bring stability
to the cities.
7. REFERENCES
[1] Durlacher Research Ltd. “Mobile Commerce Report”,
February, 2000.
[2] O. Granstrand. “Temporal Diffusion and Population
Dynamics : A Systems Model”, in. N. Nakicenovic and A.
Grubler (eds).Diffusion of Technologies and Social
Behavior. 1991. Springer-Verlag Lecture Notes in Computer
Science.
[3] J. Sairamesh and J. Kephart, “Price Dynamics in Vertically
Differentiated Information Markets”, proceedings of the
ICE-98 conference, and Springer Verlag Lecture Notes in
Computer Science, 1999.
[4] S. Wasserman and K. Faust, “Social Network Analysis”,
Cambridge Press, 1999.
[5] D. J. Watts, “Small Worlds”, 1999, Princeton University
Press.
[6] P. Krugman. The self-organization economy. 1996.
Blackwell Publishers, Inc.
[7] M. Fujita, P. Krugman and A. Venables, “The spatial
economy: Cities, regions and international trade”, 1999. MIT
Press.
[8] P. Kavassalis et. al., “Regularities in the formation and
evolution of Information Cities”, submitted to the
Symposium on Digital Cities, Kyoto, Japan, 2001.
[9] J. Jacobs, “The Economy of Cities”, Vintage Books edition,
published by Random House Edition, 1970.
[10] S. Graham, “Cyberpsace and the City”, 1995, Journal of
Town and Planning.
[11] S. Graham and A. Aurigi, “Virtual Cities, Social Polarization
and Crisis in the Urban Public Space,” April 1997, Journal
of Urban Technology.
[12] T. Ishida, “Understanding Digital Cities”, in Digital Cities:
Technologies Experiences and Future Perspectives, 2000,
Springer Verlag Lecture Notes in Computer Science.
[13] W. Mitchell, “e-Topia”, MIT Press, 1999.
[14] W. Mitchell, “City of Bits”, MIT Press, 1996.
[15] T. Ishida, Community Computing and Support Systems,
Springer-Verlag, Lecture notes in Computer Science ,1997.
[16] Virtual Amsterdam, www.ddl.nl
[17] Official New York, www.nyc.gov
[18] Boston Online, www.boston.com
[19] Yahoo Inc, www.yahoo.com
[20] America Online, AOL and Time-Warner, www.aol.com
[21] VerticalNet Inc, www.verticalnet.com
[22] Information Cities Project, Project no: IST-199911337,
Brussels, Belgium. URL: http://icities.csd.uoc.gr.
[23] S. Haridi et. al., Programming Languages for Distributed
Applications, New Generation Computing, 3: 223-261,
Omsha Ltd. And Springer-Verlag, 1998.
[24] J. Kephart et. al., “Information Filtering Economy”,
proceedings of the ICMAS’98.
[25] Y. Jung and A. Lee. Design of a Social Interaction
Environment for Electronic Marketplaces. In Proceedings of
DIS'2000 - Designing Interactive Systems: Processes,
Practices, Methods, Techniques, ACM, pp. 129-136, 2000.