You are on page 1of 7

Digital Ecosystems: Principles and

Semantics *

Boley, H., and Chang, E.


February 2007

* published at the 2007 Inaugural IEEE International Conference on Digital


Ecosystems and Technologies. Cairns, Australia. February 2007. NRC
48813.

Copyright 2007 by
National Research Council of Canada

Permission is granted to quote short excerpts and to reproduce figures and tables
from this report, provided that the source of such material is fully acknowledged.
Digital Ecosystems: Principles and Semantics
Harold Boley¹, Elizabeth Chang²

¹Institute for Information Technology - e-Business, National Research Council of Canada. Frederiction, Canada
²Digital Ecosystems and Business Intelligence Institute, Curtin University of Technology, Western Australia
e-mails:harold.boley AT nrc.gc.ca; elizabeth.chang AT cbs.curtin.edu.au

Abstract Digital ecosystems transcend the generation. Members of a species are called
traditional, rigorously defined, collaborative individuals and are made up of organs. Each
environments from centralised, distributed individual can again be considered as an entire
or hybrid models into an open, flexible, ecosystem. Each organ carries out its tasks.
domain cluster, demand-driven, interactive Organs again need to interact with each other
environment. A digital ecosystem is a newly and balance each other (even though some
networked architecture and collaborative organs are more important than others). The
environment that addresses the weakness of individual supports organ collaboration and
client-server, peer-to-peer, grid, and web communication in order to achieve and
services. In this paper we provide an maintain a healthy state.
explanation of digital ecosystems, their
analogy to ecological systems, architecture, Species, which can be grouped by biological
swarm intelligence, and comparison to classification through genus, are thus
existing networked architecture. We then composed of related individuals that resemble
describe how digital ecosystems can benefit one another and are able to live together, cross-
from semantic web ontologies and rules. fertilise and interbreed. We will discuss four
Finally, we discuss issues in the essential aspects of ecosystems: interaction and
collaboration between semantically engagement, balance, domain clustered and
neighbouring digital ecosystems. loosely coupled, as well as self-organisation.
ƒ Interaction and engagement
Keywords: Digital ecosystems, swarm This refers to interspecies interaction, such as
intelligence, self-organisation, collaboration, coral polyps, tiny animals that live in colonies
semantic web and interact with nudibranchs, fish of varying
types, turtles, sea snakes, snails and molluscs.
I. The Ecosystem Analogy They live together in warm, open, clear,
shallow waters. They need to interact for
We first introduce ecosystems in the biological
social well-being and to engage with each
sense and then proceed to digital ecosystems,
other to find interesting things, and to share the
resources. Sometimes they need to unite as a
An ecosystem is a loosely coupled, domain
group to defend against threats from human
clustered environment inhabited by species,
interference, pollution or natural disaster.
each proactive and responsive regarding its
own benefit while conserving the environment. ƒ Balance
This signifies the harmony, stability and
Two intertwined types of elements create an sustainability within an ecosystem. If some
ecosystem: multiple species in an environment. species or parts of an ecosystem are getting
Species need to interact with each other and disproportionally tensioned, dried, overheated,
balance each other (even though some species or divided, the whole ecosystem may collapse:
may play a leading role at times). The ‘No benefit or gain, but pain.’ However, a
environment supports the needs of its species single point of failure need not lead to a
so they can continue generation after disaster but may become a contribution to a

1
new balance of welfare for the ecosystem as a community - ‘push-in’ rather than ‘pull-in’.
whole. For example, when coral polyps die, Many current collaborative environments are
they become a stony, branching structure as not demand driven because people are told to
part of a reef and can still provide shelter and collaborate or forced to work together, rather
maintain the balance of the reef for than enjoying collaboration arising from a
generations. perceived mutual interest of the collaborating
parties: There is a lack of consideration about
ƒ Domain clustered and loosely coupled
whether there will be a benefit or profit from
In an ecological environment, species come to
the collaboration for the collaborating entities.
an ecosystem on their own choice. They form
‘Self-organising’ refers to agents being capable
loosely coupled groups the members of which
of acting autonomously, making decisions and
share a similar culture, social habits, interests
fulfilling responsibilities. ‘Agent environment’
and objectives. Each species preserves the
is defined as an environment which contains
environment and is proactive and responsive
human individuals, information services as
for its own benefit. They are thus able to live
well as network interaction and knowledge
together and support each other for
sharing tools along with resources that help
sustainability.
maintain synergy among human beings or
ƒ Self-organisation organisations. ‘Proactive’ is defined as an
This signifies that each species is independent, entity being full of enthusiasm to participate in
self-empowered, self-prepared, undertakes team work or the community. ‘Responsive’
self-defence, is self-surviving and undertakes signifies an agent that demonstrates
self co-ordination through swarm intelligence. willingness, is cooperative and takes
In case of natural disaster they cannot ask responsibility for its action. ‘Benefit’ refers to
‘where is the president’, ‘what logistics an advantage that an agent can take without
systems are provided’ and so forth. any risks. ‘Profit’ refers to social and
economic gain.
We propose, by analogy to such biological
ecosystems, that a digital ecosystem be defined II. The Basis of Digital Ecosystems
as an open, loosely coupled, domain clustered,
demand-driven, self-organising agent A Digital Ecosystem is: unlike a client-server
environment, where each agent of each species architecture, where the communication is
is proactive and responsive regarding its own centralised and which acts as a command and
benefit/profit (as detailed below) but is also control environment; unlike a Peer-to-Peer
responsible to its system. architecture, where, at any time, each agent
has a well defined role, i.e. can only be client
‘Agents’ are entities that join an environment or server, but not both; unlike a Grid
or a community based on their own interests. architecture, which stitches partners together
‘Species’ are types of agents. ‘Open’ refers to for resource sharing but cannot avoid counter-
a transparent virtual environment. ‘Loosely free riding; unlike a Web service network,
coupled’ refers to a freely joined, open where brokers are centralised and service
relationship between agents or species within a requesters and providers are distributed in a
virtual community. This term is in contrast to hybrid architecture that does not guarantee
a tightly coupled relationship where each party trust and QoS. A Digital Ecosystem instead is
is heavily dependent on the other, and the roles an open community, and there is no permanent
are pre-defined. ‘Domain clustered’ is the need for centralised or distributed control or
characteristic of a colony or a field where for single-role behaviour. In a Digital
species share the same life or interests such as Ecosystem, a leadership structure may be
an ocean habitat of a coral reef or exotic formed (and dissolved) in response to the
tropical plants in a rainforest. ‘Demand dynamic needs of the environment.
driven’ is defined as the driving force to join a

2
An agent in a Digital Ecosystem can be a client not just one way. (2) Leading agent and
and a server at the same time. In the same species: They emerge through (temporary)
message, agents may offer a service to others hierarchy formation to facilitate, lead and
as a Server and request help as a Client. There direct collaborative swarms; they may be the
is no centralised control structure or fixed role representative of the domain cluster in the
assignment. There is no preconfigured global interaction with other ecosystems. They have
architecture, where the communication and the same features and functions as any other
collaboration is based on swarm intelligence: agent, but in the current situation have
Unlike traditional environments, digital activated the general leadership potential.
ecosystems are self-organising systems which
can form different architectural models Animals, humans, software agents or
through swarm intelligence, where local autonomous robots can all be analysed as
interactions between agents determine the agents in the above sense. In this analogy with
global behaviour. Occasionally, intelligent biological (and social) ecosystems, one may
agents or entire species may configure into a find a better basis for understanding of
hierarchical organisation where the intelligence and rationality than that provided
communication channels are defined with a by traditional AI. Swarm intelligence can help
leading agent (cf. leading bird or queen bee).. us model intelligent behaviour in relation to,
Some intelligent agents or species institute a e.g., rational behaviour, goal seeking, task
workflow process with sequentially ordered accomplishment, and learning.
tasks and predefined flow of operations. Other
intelligent agents or species collaborate in a III. Semantics for Digital Ecosystems
control loop, where each agent is self-
coordinated and they put their energy together Individuals in ecosystems need to exchange
to tackle issues iteratively. and process messages to coordinate their
behaviour. The information that individuals
A swarm is a set of agents which have send to other individuals or broadcast to an
common characteristics and are able to interact entire (sub)system reveals its meaning or
and engage directly or indirectly with each semantics by a process of interpretation in the
other. They collectively carry out a task or receiving individuals. Biological ecosystems
share a problem. Swarm intelligence is an have developed shared implicit semantics, step
important property of ecosystems. We often by step, during their long evolution. Social
observe the collective behaviour of agents or ecosystems have enriched this with shared
species interacting with each other and with explicit informal semantics (communicated via
the environment, and generate a coherent natural languages) mainly through behavioural
functional global pattern. Swarm intelligence and linguistic conventions, regulations, and
is now widely researched as it provides a basis laws. Digital ecosystems should add shared
to explore collective behaviour for problem explicit formal semantics (communicated via
solving without centralised or command and artificial languages) to enable automation with
control systems, and the provision for flexible, high precision in several areas of business,
dynamic interactive models. government and other domains.

Let us consider here two aspects of swarm Digital ecosystems can benefit from ongoing
intelligence. (1) Agent and species: They are work in the semantic web, where each entity
the foundation of intelligence; they can be (e.g., a document) can be globally identified
viewed as an individual or an organisation. with an Internationalized Resource Identifier
Each has its own niche or role to play; each has (IRI). These IRIs can then be used as reference
dual functions or roles. They can be client and points of (globally distributed) semantic
can also be server at the same time; each one metadata that enable to search entities with
can carry out bi-directional communication, high precision. Metadata (e.g., written in the

3
RDF language) are descriptions used for entity rely on referrals by other experts using their
indexing by categorizing an IRI into classes social networks. FindXpRT thus provides both
and attaching properties to them, so that an IRI direct searches and referrals, which are both
(and the entity it identifies) can be found accomplished by applying rules to users' expert
through a kind of ‘associative’ retrieval by queries. We also propose a benchmark suite for
querying with a subset of descriptive classes expert finding more generally, testing expert-
and properties. High precision of such retrieval finding systems against expert profiles. This is
is enabled by an ontology (often modularised exemplified with our implemented system,
into subontologies) which acts as the agreed- tested against the expertise and co-expertise
upon vocabulary of a digital ecosystem. The domains of computer science and music,
ontology (e.g., written in the RDFS or OWL respectively.
language) groups classes (and properties) into
a hierarchy, and specifies which (domain) IV. Collaboration Between Digital
classes can have which properties of which Ecosystems
(range) classes. Furthermore, rules (e.g.,
written in the RuleML language) can be used The semantics get more complex and
to derive (virtual) properties from interesting when a given individual or
combinations of other (stored) properties, to (sub)system can be part of multiple
check the integrity of the metadata, and also to overlapping ecosystems, hence interact with
process the retrieved information. individuals in any one of them (perhaps at
different times). Biological ecosystems often
The entities described and queried in this develop pairwise overlaps in geographic
highly precise, semantic way, can be arbitrary borderline regions such as the outskirts of a
documents, multimedia objects, web services, forest or the tidelands of an ocean. Social
etc. that are of use within a digital ecosystem. ecosystems typically overlap more freely in
However, these entities can also be agents or various ways such as groups formed around,
entire (sub)systems, because metadata can act say, family, profession, and hobby. Digital
as profiles describing the capabilities of agents, ecosystems may overlap like social ones, but
human or machine, for social or digital entirely remove the limitations of geographic
networking. This means that an agent of a proximity and can provide tools for cross-
(large) digital ecosystem can semantically system collaboration.
search for other agents that advertised
themselves using such metadata profiles and The single-ecosystem case in previous sections
that fall into certain classes and fulfil certain can be generalised to this multi-ecosystem
properties. Such agent profiles are often case. In particular, ‘the’ semantic web
written in RDF, using an RDFS-based explained in section III actually is divided into
ontology known as FOAF or the recently multiple semantic webs serving various digital
proposed SIOC. While existing profiles are ecosystems. Even in a ‘single’ field of
fact-based, rules can describe properties knowledge such as medicine there can be
conditional on other agents, the time, the ‘multiple’ semantics such as those of
location, and so on. traditional and orthodox medicine (and even
for orthodox medicine there exist multiple
We have implemented the FindXpRT system ontologies such as SNOMED and UMLS).
that applies rule-based social or digital Medicine also overlaps with neighbouring
networking to expert finding. The assumed fields such as chemistry, which is reflected by
business-service model is the bartering-like their partially shared digital ontologies and
exchange of expertise between an expert and a rules. Sometimes such overlaps develop into
co-expert, where the latter initiates the search fields of their own such as the medicine-
using our system. When searching for an chemistry overlap of pharmacology.
expert, in any domain, humans often need to

4
Participation of individuals in multiple digital to bridge between research in digital ecosystem
ecosystems can obviously help them to reach and the semantic web. Government and
their goals. Collaboration between entire business application of ‘semantic ecosystems’
digital ecosystems can also provide benefits to should contribute to international productivity,
all ecosystems involved. First, the current prosperity and social, cultural and economic
borderlines between semantically neighbouring balance as well as ecological sustainability.
digital ecosystems may not be optimal for any Digital Ecosystems thus move from Darwinian
system involved or may indeed be counter- competition to self-interested collaboration.
productive, so redrawing or even removing
(some of) them may be advantageous to all References
these ecosystems. The joining of an agent
(viewed as a singleton ecosystem) into a digital 1. http://www.w3.org/Collaboration
ecosystem can be generalised to the merger of 2. http://www.w3.org/International/O-URL-and-
two digital ecosystems. Second, while a ident.html
number of digital ecosystems may keep their 3. http://www.edgelab.ca/CSCW/Workshop2004/
borders and identities, they might specialise on WorkshopNotes.pdf
4. Alston, J., Hess, D. & Ruggieo, R., 2002,
what they can do best and form
'UDDI (Universal Description, Discovery, and
‘import/export’-connected components of a Integration) and web services: Perspective',
higher system as well-known from biological Gartner.
and social ecosystems. 5. Berman, F., Fox, G. & Hey, A., 2003,
'Overview of the book: Grid computing -
Semantic support for digital collaboration has making the global infrastructure a reality', in
been studied in the context of semantic web Berman, F., Hey, A. & Fox, G. (eds.), 'Grid
and web service projects under headings such computing - making the global infrastructure a
as “semantic information intergration”. Since, reality', John Wiley & Sons, Ltd.
at least at the start of a collaboration, the 6. Boley, H. The RuleML Family of Web Rule
Languages, Invited Talk, Proc. 4th Workshop
ontologies and rules of component systems are
on Principles and Practice of Semantic Web
usually not (fully) compatible, alignment and Reasoning, Budva, Montenegro, June 10-11,
mapping/translation techniques have been 2006, pp. 1-15, Springer LNCS 4187.
studied. For ontologies these include 7. Carrol, A., Juarez, M., Polk, J. & Leinger, T.,
Chimaera, PROMPT, and RDFT. For rules, 2002, ‘Microsoft palladium: A business
interoperability techniques are being overview', Microsoft whitepaper, June 2002.
developed, e.g. by the standard bodies OASIS, http://www.microsoft.com/PressPass/features/2
OMG, and W3C. In particular, like the 002/jul02/0724palladiumwp.asp.
RuleML Initiative, W3C’s RIF Working Group 8. Cerami, E., 2002, 'Web services essentials',
is dedicated to XML- and RDF-based rule O'Reilly & Associates, Inc., Sebastopol.
9. Chang, E., Dillon, T.S. & Hussain, F., 2005,
interchange via translators to and fro RIF,
‘Trust and reputation for service oriented
written in languages such as XSLT. environments- Technologies for building
business intelligence and consumer
V. Conclusion confidence’, John Wiley & Sons, ISBN: 0-470-
01547-0
This paper provides an explanation of digital 10. Cruz, S. M. S., Campos, M. L. M., Pires, P. F.
ecosystems, their architecture and comparison & Campos, L. M., 2004, 'Monitoring e-
with most advanced communication platforms business web services usage through a log
or environments such as client-server, P2P and based architecture', Proceedings of the IEEE
web services. It also describes how digital International Conference on Web Services
(ICWS'04).
ecosystems can benefit from semantic web
11. Denning, P.J. & Metcalfe, R.M., 1998, 'Beyond
ontologies and rules, and discusses calculation: The next fifty years of computing',
collaboration issues between semantically (edited) Springer, ISBN: 0387985883
neighbouring digital ecosystems. We thus try

5
12. Duong, D.D., Boley, H., Le, T.T.T., Bhavsar, 24. Novak, B., “Introduction to RPDE Program”
V.C., “Rulebase Integration for DEBI Seminar, Curtin University, 16 August
eCollaboration,” in ‘e-Networks in an 2006.
Increasingly Volatile World,’ A. Gaudes (ed.), 25. Parameswaran, M., Susarla, A. & Whinston, A.
Proc. of the 11th International Workshop on B., 2001, 'P2P networking: An information-
Telework, Fredericton, NB, Aug. 28-31, 2006, sharing alternative',
ISBN 1-55131-107-0, pp. 2-17, August 2006. http://crec.mccombs.utexas.edu/works/articles.
13. European Commission, 2006, ‘Innovation 26. Roure, D. D., Baker, M. A., Jennings, N. R. &
ecosystems – A specific roadmap’, DG Shadbolt, N. R., 2003, 'The evolution of the
Information Society and Media. grid', in Berman, F., Hey, A. & Fox, G. (eds.),
14. European Commission, 2006, ‘Technologies 'Grid Computing - Making the Global
for digital ecosystems – Innovation ecosystem Infrastructure a Reality', John Wiley & Sons,
initiative’, available at http://www.digital– Ltd.
ecosystems.org 27. Roy, J. & Ramanujan, A., 2001,
15. Engelbrecht, A.P., Fundamentals of 'Understanding web services', IT Professional,
Computational Swarm Intelligence, John vol. 3, no. 6, pp. 69-73.
Wiley & Sons, 2005. 28. Samtani, G. & Sadhwani, D., 2002, 'Web
16. Fishman, M., et al., 2000, ‘Valuating services and peer-to-peer computing', Web
community collaborations: A research services Architect.
synthesis’, Lewin Group, Inc, USA. http://www.webservicesarchitect.com/content/
17. Foster, I., Kesselman, C., Nick, J. M. & 29. Schneider, J., 2001, 'Convergence of peer and
Tuecke, S., 2003, 'The physiology of the grid', web services', Available:
in Berman, F., Hey, A. & Fox, G. (eds.), 'Grid [http://www.openp2p.com/lpt/a/1047] (Oct 6,
computing - Making the global infrastructure a 2004).
reality', John Wiley & Sons, Ltd, pp. 217-249. 30. Toh, C. K., 2001, 'Ad hoc mobile wireless
18. Gannon, D., Ananthakrishnan, R., Krishnan, networks: Protocols and systems', Prentice Hall
S., Govindaraju, M., Ramakrishnan, L. & PTR.
Slominski, A., 2003, 'Grid web services and 31. Tomoya, K. & Shigeki, Y., 2003, 'Application
application factories', in Berman, F., Hey, A. & of P2P (peer-to-peer) technology to marketing',
Fox, G. (eds.), 'Grid computing - Making the Proceedings of the 2003 International
global infrastructure a reality', John Wiley & Conference on Cyberworlds (CW’03), pp. 1-9.
Sons, Ltd, pp. 251-264. 32. Tsaparas, P., 2004, 'P2P search', Available:
19. Hess, D., 2002, 'Simple object access protocol [www.cs.unibo.it/biss2004/slides/tsaparas-
(SOAP) and web services: An introduction', myP2P.pdf] (3/10/, 2004).
Gartner. 33. Universal Description, Discovery, and
20. Huy, H. P., Kawamura, T. & Hasegawa, T., Integration (UDDI) 2000, Available:
2004, 'Web service gateway - A step forward to [http://publib.boulder.ibm.com/infocenter/wsph
e-business', Proceedings of the IEEE elp/index.jsp?topic=/com.ibm.etools.webservic
International Conference on Web Services e.consumption.doc/concepts/cuddi.htm] (Sep
(ICWS'04). 10, 2004).
21. Kifer, M., de Bruijn, Jos, Boley, H., Fensel, D., 34. W3C. Web Services Architecture
A Realistic Architecture for the Semantic Web, Requirements Oct 2002,
RuleML-2005 Conference Proc., Galway, Nov http://www.w3.org/TR/wsa-reqs.
2005, Springer LNCS 3791,17-29. 35. What is ad-hoc network (2003), Available:
22. Li, J. Boley, H., Bhavsar, V.C., Mei, J., [http://whatis.techtarget.com/definition/0,2898
“Expert Finding for eCollaboration using 93,sid9_gci213462,00.html] (7/10/2004).
FOAF with RuleML Rules,” Proc. of the 2006 36. Xiao Feng, J., Junhua, X., Hua, Z. & Zuzhao,
Conference on eTechnologies, (Eds.) L. L., 2003, 'A realizable intelligent agent model
Logrippo, H. Mili and A. Salah, Montreal, applied in dynamic e-business'. Proceedings of
Canada, May 2006, pp. 53-65, 2006. the IRI 2003. IEEE International Conference
23. Napster, http://www.napster.com/ntg.html on Information Reuse and Integration,
http://ieeexplore.ieee.org/xpl