Inflab.kard.teilar.gr

On Self-Organising Mechanisms from Social, Business and Economic Domains Salima HassasLIRIS-CNRS, University of Lyon, FranceE-mail: hassas@liris.cnrs.frhttp://www710.univ-lyon1.fr/~hassas Giovanna Di Marzo-SerugendoUniversity of Geneva, SwitzerlandE-mail: Giovanna.Dimarzo@cui.unige.chhttp://cui.unige.ch/~dimarzo/ Anthony KarageorgosUniversity of Thessaly, GreeceE-mail: karageorgos@computer.orghttp://inf-server.inf.uth.gr/~karageorgos/ Cristiano CastelfranchiUnit of AI, Cognitive Modelling and Interaction, CNR, ItalyE-mail: cristiano.castelfranchi@istc.cnr.ithttp://www.istc.cnr.it/ Keywords: self-organisation, networks, social functions, business networks, social learning This paper discusses examples of socially inspired self-organisation approaches and their use to buildsocially-aware, self-organising computing systems. The paper presents different mechanisms originatingfrom existing social systems, such as stigmergy from social insects behaviours, epidemic spreading, gos-siping, trust and reputation inspired by human social behaviours, as well as other approaches from socialscience related to business and economics. It also elaborates on issues related to social network dynamics,social network patterns, social networks analysis, and their relation to the process of self-organisation. Theapplicability of socially inspired approaches in the engineering of self-organising computing systems isthen illustrated with applications concerning WWW, computer networks and business communities.
Povzetek: Podani so primeri mehanizmov samoorganizacije.
uncertain and dynamic environments. They can providea great inspiration for busiding self-organising computingsystems.
Nowadays computing systems are open systems evolvingin a dynamic complex environment. They are designed as Socially inspired computing gathers computing tech- sets of interacting components, highly distributed both con- niques that make use of metaphors inspired by social be- ceptually and physically. The growing complexity of these haviours, exhibiting self-organisation, self-adaptation and systems and their large scale distribution make the use of self-maintainance of the society organisation. These so- traditional approaches based on hierarchical functional de- cial behaviours range from those observed in biological composition and centralised control no more applicable.
entities such as bacteria, cells and social insects to an- Increasingly, a real need for new paradigms, mechanisms imals and human societies. One important characteris- and techniques allowing endowing these systems with the tic of these societies is their emergence as patterns de- capacity to autonomously manage their functioning and veloped from relatively simple interactions in a network evolution, is expressed. Existing social systems, for exam- of individuals. These patterns, are supposed to be driven ple large scale, decentralised and autonomic human, insect by self-organising processes that are governed by sim- or business and economic systems, are well known to ex- ple but generic laws [19][5]. This paper is focused on hibit interesting characteristics, such as robustness, capac- self-organising mechanisms observed in natural social sys- ity of self-management and self-adaptation, survivability in tems and in business and economic ones, and the illus- tration of their use for building self-organising computing cial emergence: 1. the emergent phenomenon is perceived systems. We distinguish natural systems from business by an observer, but has no effect on the society; 2. the and economic systems, since generic laws guiding self- emergent phenomenon has an effect on the society by self- organisation in the first kind of systems is dictated by na- reproducing and enforcing the social phenomenon.
ture whereas in the others, self-organisation is governed by Given the considerations above, Castelfranchi considers that "in order to have a function, a behaviour or trait or From a natural systems perspective, species survival is entity must be replicated and shaped by its effects".
the ultimate goal. This goal is not expressed explicitly The principal argument is that "the invisible hand" is not at the individual level, but seems to guide the collective necessarily a good thing for society (especially in the case behaviour towards the emergence of social functions and of self-interested agents). The optimum order for the so- dynamics allowing the maintainance of the system organ- ciety can actually be bad for individuals or for everybody.
isation. In business and economic systems, individual be- For instance, prisons generate criminals that in turn feed haviours are goal-oriented and their primary goal is to in- prisons. This is a function not a social objective.
crease their profit. In this case, the system’s dynamics is The important thing is that "re-organisation simply handled by the activity developed to face business and eco- maintains the system, but not necessariy the optimal nomic constraints to reach a global equilibrium through which the system can survive. In both systems, one im-portant issue is their capacity to globally maintain a suffi-ciently good level of information allowing them to deploy the effective global behaviour that permits the realisation of their intentional or non intentional goals.
In the following, we first present examples of socially in- spired self-organising mechanisms in natural business and Propagation of information or knowledge allowing social economic systems. Before concluding, we present exam- activities in social systems lays on the social network ple applications of such mechanisms in WWW, computer formed by the the interaction held between the society in- dividual components during social activities. Social be-haviour both shapes and is shaped by such social networks.
2.2.1 Social Learning and Propagation of Knowledge In social science, it is now established that social interac- tions play a fundamental role in learning dynamics, andlead to cognitive development. This phenomenon is known Human collective behaviour occurs without central con- as "Zone of Proximal Development" which Vygotsky de- trol, and through self-organisation. In this case, intimately scribes it as "the distance between the actual development linked with the notion of self-organisation is the notion of level as determined by independent problem solving and "emergence" in the sense that "social functions" arise out the level of potential development as determined through from (self-interested) human collective behaviour. In so- problem solving under adult guidance or in collaboration cial sciences different interpretations of the notion of social with more capable peers" [51] [15]. The effect of social- functions have been expressed, essentially considering that isation has also been proven to benefit to the propagation even if social functions are not intentional and possibly un- of knowledge inside an interconnected population. In [14] known they constitute the ultimate end of the society and the authors considered social learning in a population of myopic, memoryless agents. They have made some exper- The social functions concept has also been explained iments to study how technology diffuses in a population as the "invisible hand" which would manage forms of un- based on individual or collective evaluation of the tech- planned coordination (like market) in which human interest nology. The authors have shown that under a learning increases [31] through the apparently "spontaneous emer- rule where an agent changes his technology only if he has gence of an unintentional social order and institutions". As had a failure (a bad outcome), the society converges with pointed out by [13], the problem with this view is: "how probability 1 to the better technology. In contrast, when an unintentional effect can be an end" for the society; and agents switch on the basis of the neighbourhood averages, "how is it possible that we pursue something that is not an convergence occurs if the better technology is sufficiently intention of ours". An alternative could be avoiding the better. These experiments show how a better technology concept of social functions because of the problems and spreads in a population through a mechanism of imitation questions that they provoke. However, this is not satisfac- and thanks to neighbourhood connections. In another work tory too, because nevertheless social emergence happens [3], the authors develop a general framework to study the and has the form of a goal-oriented process.
relationship between the structure of these neighborhoods Therefore, it is important to distinguish two kinds of so- and the process of social learning. They show that, in a ON SELF-ORGANISING MECHANISMS FROM SOCIAL. . .
connected society, local learning ensures that all agents ob- of evidence, and allows to adapt the behaviour of princi- tain the same payoffs in the long run. Thus, if actions have pals consequently. We report here the results of the Euro- different payoffs, then all agents choose the same action.
pean funded SECURE [11] project, which has establishedan operational model for trust-based access control. Sys- 2.2.2 Epidemic Spreading and Gossiping Metaphors tems considered by the SECURE project are composed ofa set of autonomous components, called principals, able to As cited in [34] Gossip is one of the most usual social take decisions and initiatives, and are meaningful to trust or activities. This mechanism allows for the aggregation of distrust. Principals maintain local trust values about other a global information inside a population, through a peri- principals. A principal that receives a request for collabora- odic exchange and update of individual information among tion from another principal decides to actually interact with members of a group. The neighbourhood as well as the that principal or not on the basis of the current trust value level of precision of the exchanged information play an it has on that principal for that particular action, and on the important role on the nature of social learning which oc- risk it may imply for performing it. If the trust value is too curs by this way. This mechanism provides a powerful ab- low, or the associated risk too high, a principal may reject straction metaphor for information spreading, knowledge the request. After each interaction, participants update the exchange and group organisation in large scale distributed trust value they have in the partner, based on the evaluated systems. In peer-to-peer (P2P) systems, a class of proto- outcome (good or bad) of the interaction. A principal may cols categorised as epidemic protocol has been proposed also ask or receive recommendations (in the form of trust [50]. These protocols are characterised by their high ro- values) about other principals. These recommendations are bustness and large scalability. This metaphor has been also evaluated (they depend on the trust in the recommender), used for routing in sensor networks. For example in [8], a and serve for updating current trust values. Artificial sys- rumour routing algorithm for sensors networks is proposed.
tems built on the human notion trust as exposed above have This algorithm is based on the idea of creating paths lead- the particularity to exhibit a self-organising behaviour [16], ing to each event and spreading events in the wide-network as identified by Nobel prize Ilya Prigogine and his col- through the creation of an event flooding gradient field. A leagues [24]. Additional trust and reputation systems are random walk exploration permits to find event paths when surveyed in [25], and for the particular case of multi-agent Uncertainty and partial knowledge are a key characteris-tic of the natural world. Despite this uncertainty human beings make choices, take decisions, learn by experience,and adapt their behaviour.
Social insects societies such as ants, bees, wasps and ter- Trust management systems deal with security policies, mites exhibit many interesting complex behaviours such as credentials and trust relationships, for example issuers of emergent properties from local interactions between ele- credentials. Most trust-based management systems com- mentary behaviours achieved individually. The emergent bine higher-order logic with a proof brought by a requester collective behaviour is the outcome of a process of self- that is checked at run-time. These systems are essentially organisation, in which insects are engaged through their based on delegation, and serve to authenticate and give repeated actions and interactions with their evolving en- access control to a requester [53]. Usually the requester vironment [32]. Self-organisation in social insects relies brings the proof that a trusted third entity asserts that it on an underlying mechanism : Stigmergy, originally in- is trustable or it can be granted access. These techniques troduced by Grassé in 1959 [26]. Grassé studied the be- have been designed for static systems, where an untrusted haviour of a kind of termites during the construction of client performs some access control request to some trusted their nests and noticed that the behavior of workers during server [1, 6]. Similar systems for open distributed environ- the construction process is influenced by the structure of ment have also been realised, for instance [38] proposes the constructions themselves. This mechanism is a power- a delegation logic including negative evidence, and dele- ful principle of cooperation in insect societies. It has been gation depth, as well as a proof of compliance for both observed within many insect societies such as wasps, bees parties involved in an interaction. The PolicyMaker sys- and ants. It is based on the use of the environment as a tem is a decentralised trust management systems [4] based medium of inscription of past behaviours effects, to influ- on proof checking of credentials allowing entities to locally ence future behaviours. This mechanisms defines what is decide whether or not to accept credentials (without relying called auto-catalytic process, that is the more a process oc- to a centralised certifying authority). Eigentrust [36] is a curs, the more it has a chance to occur in the future. More trust calculation algotrithm that allows to calculate a global generally, this mechanism shows how simple systems can emergent reputation from locally maintained trust values.
produce a wide range of more complex coordinated behav- Recently, more dynamic and adaptive schemas have been iors, simply by exploiting the influence of the environment.
defined, which allow trust to evolve with time as a result Many behaviours in social insects, such as foraging or col- lective sorting are rooted on the stigmergy mechanism.
modelling self-organisation and emergence in economic Foraging is the collective behaviour through which ants systems, which is primarily based on analytic general equi- collect food. During the foraging process, ants leave their librium models, for example as is done in [22]. The main nest and explore their environment following a random problem with analytic approaches is that they cannot rep- path. When an ant finds a source of food, it carries a piece resent all possible situations due to the non-linearity of of food and returns back to the nest by laying a trail of economic phenomena [10], which is due to the fact that a hormone called pheromone along its route. This chem- economies are complex dynamic systems [48]. Instead, ical substance persists in the environment for a particular market-based approaches view macroeconomic phenom- amount of time before it evaporates. When other ants en- ena as emergent results of local interactions of the eco- counter a trail of pheromone, while exploring their environ- nomic entities [10, 33, 48]. An example is economic ment, they are influenced to follow the trail until the food growth which can be described at the macro level but it source, and while coming back to the nest they enforce the can never be explained at that level [12]. The reason is that initial trail by depositing additional amounts of pheromone.
economic growth results from the interaction of a variety The more the trail is followed, the more it is enforced and of economic actors, who create and use technology, and has a chance to followed by other ants in the future. Ants foraging behaviour have inspired many works in comput- There are numerous variations of market-based self- ing domains, ranging from "Ant Colony Optimisation" (ACO) organisation mechanisms. An exemplar such mechanism metaheuristic for optimisation problems [18], to the de- which is based on the creative destruction principle is de- sign of ant-like systems using mobile agents with applica- tions in several domains such as computers network routingand load-balancing [42][17][21], computers network secu-rity [20][23], information sharing in peer to peer systems Creative destruction is a term coined by Schumpeter [43] Collective clustering and sorting is a collective be- to denote a "process of industrial mutation that incessantly haviour through which some social insects sort eggs, lar- revolutionizes the economic structure from within, inces- vae and cocoons. As mentioned in [7], an ordering phe- santly destroying the old one, incessantly creating a new nomenon is observed in some species of ants when bodies one." In other words, creative destruction occurs when a are collected and later dropped in some area. The proba- new setting eliminates an old one leading to economic de- bility of picking up an item is correlated with the density velopment. According to this view an economic system of items in the region where the operation occurs. This be- must destroy less efficient firms in order to make room haviour has been studied in robotics through simulations for new, possibly more efficient entrants. A representa- and real implementations [32]. Robots with primitive be- tive example of creative destruction is the evolution of per- haviour are able to achieve a spatial environment structur- sonal computer industry which under the lead of Microsoft ing by forming clusters of similar objects via the mecha- and Intel destroyed many mainframe computer companies; nism of stigmergy described above. Moreover, these kind however, at the same time one of the most important tech- of social insect behaviours have inspired many mechanisms nological achievements of this century was created.
for building artificial self-organised systems [7][32] [30] The main roles that economic actors play in a market- based economy are those of producer, worker and con-sumer. Producers produce goods or provide services that consumers demand. Consumers consume the goods anduse the services in exchange of some monetary or utility value. When there is high demand producers tend to hireworkers to assist them in goods production or service pro- vision in exchange of some wage. Since producers cannotsell their production beforehand, they must hold enough Market-based mechanisms are built along the lines of eco- money to pay the workers in order to start up production nomic markets. In this approach, systems are modelled and they can only get the necessary money by entering along the lines of some economic model in which partic- debt. According to the creative destruction principle, if ipating entities act towards increasing their personal profit producers are not able to pay the worker wages then they or utility. System wide parameters are modelled in a man- go bankrupt and they are removed from the system, for ex- ner similar to macroeconomic variables such as economic ample they are reduced to simple workers, opening the way growth. The parameters of the individual entities corre- to other economic entities to try to become successful pro- spond to microeconomic parameters. The key point in such ducers and satisfy the consumer demand.
systems is to select suitable micro level parameter values The creative destruction process is better illustrated in a and market interaction rules so that desired system goals, credit economy. In contrast to a monetary economy where both local and global, are achieved.
producers can only borrow existing money from lenders, Market-based approaches contrast the traditional way of credit economy allows producers to obtain credit up to a ON SELF-ORGANISING MECHANISMS FROM SOCIAL. . .
certain level from creditors in order to pay for production A shift towards to personalised marketing models is of new products. In this way, producers can more easily viewed as being driven by syndication [54]. Syndication force their way into the market but the danger of becom- involves the sale of the same good to many customers, who ing bankrupt is increased. To explain economic develop- then integrate it with other offerings and redistribute it, as ment in this framework one only needs to explain why en- is the case in redistributing popular TV programs. An ex- trepreneurs would want to introduce new products to the ample of a company using syndication is FedEx which syn- market. Effective entrepreneurs survive the battle and in- dicates its tracking system in several ways [54]. The com- crease their profit. Failed entrepreneurs cannot repay their pany allows customers to access computer systems via its debt and therefore they go bankrupt and they are elimi- Web site and monitor the status of their packages. For cor- nated. As initially stated by Schumpeter [43] and later eval- porate customers FedEx provide software tools that enable uated experimentally, for example [9], economic growth in the organisation to automate shipping and track packages this model is generated in cycles that emerge from the dis- using their own computing resource. Each customer is of- turbance caused by entrepreneurs entering the market in- fered different prices depending on a variety of parameters.
Many websites, such as eBay, also apply variable pricing In such a model there is particular interest from both the global, macro economic perspective and the local mi-croeconomic one.
their entrepreneur policy so that to increase their profit and avoid the risk of getting bankrupt. On the other hand Another example from the area of management is the the- the economic system regulators can decide on the self- ory of activity described in [49]. In this view a company organisation rules so that to increase overall system pro- consists of networks of working groups that can change their structure, links and behaviour in response to businessrequirements. The aim is to capture the self-organisation decisions that need to be taken during the business oper-ations both by managers and by interactions between em- Business related mechanisms are based on business models ployees. The emphasis is on solving potential conflicts of and theories which use self-organisation. In an increasingly interests in both the inner and the external co-operative ac- complex global economy, businesses are faced with unpre- dictable behaviours and fast pace of change. As a result, In this approach the structure of the company is virtual.
the emphasis in contemporary business models has shifted There is no clear hierarchy and control; instead control ef- from efficiency to flexibility and the speed of adaptation.
fects can be initiated both vertically and horizontally via More recent approaches, for example the one described "round table meetings", which are organised along the lines in [46], increasingly introduce business models originat- of assessment meetings normally held in companies to as- ing from the study of complex adaptive systems. Adap- sess results and handle exceptions. In these virtual round tive business organisations are guided and tied together by tables suitable participants soon emerge as de facto leaders ideas, by their knowledge of themselves, and by what they due to their knowledge and experience. Subsequently, lead- do and can accomplish. Therefore, the focus in such mod- ers tend to participate in each newly formed "round table".
els is on the complex relationships between different busi- The view expressed in [49] is that to model the interactions ness components and the effects that a change into some of participants in a "round table", it is necessary to simulate part of the system or its environment, however distant, the whole activity of each of them including their reasoning might have on the behaviour of the entire system.
As examples of self-organising business models we dis- cuss personalised marketing and activity-based manage-ment.
Personalised marketing refers to following a personalised market strategy for each individual customer which is evolving according to customer reactions [52]. A typicalexample of this approach is the one-to-one variable pric- Based on the SECURE trust and risk security framework, ing model [29], which refers to providing an individual an anti-spam tool has been developed which allows offer to each customer using Internet technologies. The collaboration among e-mail users by exchanging recom- model uses self-organisation in the marketing policies by mendations about e-mail’s senders.
changing customers targeted and the prices quoted based scheme has been combined to the SECURE framework in on market dynamics, customer characteristics and the busi- order to increase the level of sender authentication [44].
On the WWW, a plethora of systems have been devel- 5.3 Applications in Business and Economics oped for content retrieval, filtering or organisation using socially inspired computing. As an illustration, we presenthere a pioneering work [40], in information retrieval field Typical applications of market-based self-organisation which combined inspiration of social human behaviours, mechanisms can be found in the domains of business com- and economic markets to propose an interesting system for munity networks [37]. An example of such approach is information retrieval on the web. In this work, documents the self-organising semantic network of document index- are represented by keyword vectors, representing individ- uals (agents) of an artificial ecosystem. This population In such a network, agents maintain indices to actual doc- evolves through an evolutionary process of natural selec- uments and to other agents as well, treating both in a similar tion using a genetic algorithm to find documents which manner - based on the semantics of their content. The key best fit the user request. The user feedback is used to re- feature in this approach is content dependent query redirec- ward (resp. to punish) the fittest individual (the less fitting tion, based on semantic indexing. If an agent is unable find individual) by giving it a credit value. These credits are a document on a given topic, it re-directs the received query then used by agents in a market based metaphor to esti- to the agents which believes are most likely to find it. The mate the cost of inhabiting the artificial ecosystem. The connections between the agents adapt themselves based on fittest agents have enough credits to continue living in the the history of successfully served queries, forming a dis- ecosystem and the less fitting agents will die. Another sys- tributed self-organising search engine which is capable of tem called WACO has been proposed in [30]. The WACO executing on heterogeneous servers over the internet and system is composed of a population of agents deployed on dynamically indexing all available documents. The impor- the web to form clusters of semantically similar documents tant aspect of such a search engine is that each node, though and dynamically organise the web content. These agent be- possessing only limited amount of local information, can haviours, take inspiration of social insect behaviours. They combine foraging ant behaviour and the collective sorting Each piece of information received from an agent cor- rects the coordinates of its representation in the semanticindex of the recipient. Furthermore, each link to an agenthas also its own utility based rating. Those ratings areused for the selction of the right candidates for redirecting Rating adaptation is done using a free market approach.
T-Man is a generic protocol based on a gossip communi- According to this approach agents provide chargeable cation model and serves to solve the topology management search services to each other. Each query has some lim- problem [35]. Each node of the network maintains its local ited amount of network currency, termed neuro, which dis- (logical) view of neighbours. A ranking function (e.g. a sipates in the course of query processing in the network.
distance function between nodes) serves to reorganise the Neuros circulating through the network are used by the set of neighbours (e.g. increasing distance). Through lo- agents to update their connections with the other agents, cal gossip messages, neighbour nodes exchange or com- based on their utility, in a similar manner that money flow bine their respective views. Gradually, in a bottom-up in a real economy determines the structure of business re- way, through gossiping and ranking, nodes adapt their list of neighbours, and consequently change and re-organise The semantic network economy is based on the follow- the network topology. The T-Man protocol is particularly suited for building robust overlay networks supporting P2Psystems, especially in the presence of a high proportion of – The cost of each delegated query processing is one nodes joining and leaving the network.
The SLAC (Selfish Link and behaviour Adaptation – The cost of each document (query) transaction is one to produce Cooperation) algorithm [28] favours self- organisation of P2P network’s nodes into tribes (i.e. intospecialised groups of nodes). The SLAC algorithm is a – Agents aim to minimize their expenditures.
selfish re-wiring protocol, where by updating its links with According to these rules each agent keeps track of the other nodes in order to increase its utility function, a spe- balance of transactions of all other agents it is linked with.
cific node leaves its current tribe, and joins a new one.
Agents are considered economically rational and aiming to In addition to P2P systems, the SLAC algorithm has maximise their profit they tend to delegate queries to ex- many potential applications, for instance to organise col- perts in the query topic, thus minimizing effective cost of laborative spam / virus filtering in which tribes of trusted peers share meta-information such as virus and spam signa- Similar market-based techniques are applied in trade net- tures. This would elimite the need for trusted third parties works where the aim is to select trade partners based on continually updated expected payoffs [27, 47].
ON SELF-ORGANISING MECHANISMS FROM SOCIAL. . .
[7] E. Bonabeau, G. Théraulaz, V. Fourcassié, and J-L.
Deneubourg. The phase-ordering kinetics of ceme- In this paper we have surveyed some self-organising social tery organization in ants. Technical Report 98-01- approaches and presented their use as metaphors for dis- tributed computing systems. These socially inspired com-puting techniques have shown their effectiveness for sys- [8] D. Braginsky and D. Estrin. Rumour routing algo- tems and applications evolving in distributed and highly rithm for sensor networks. In Proceedings of the dynamic environments, such like current complex net- Fisrt Workshop on Sensor Networks and Applications works. Social behaviours ranging from those observed (WSNA), Atlanta, GA, USA, September 2002.
in biological entities such as bacteria, cells and social in- [9] Charlotte Bruun. The economy as an agent-based sects, to animals and human societies, are rooted in the whole simulating schumpeterian dynamics. Industry dynamics of their underlying social network. Social be- and Innovation, 10(4):475–491, 2003.
haviour both shapes and is shaped by such social net-works. One important characteristic of societies is their [10] Charlotte Bruun. Introduction to agent-based com- emergence as patterns developed from relatively simple in- teractions in a network of individuals. The obtained pat- borg University, Department of Economics, Poli- terns are then enforced through dynamics underlying the tics and Public Administration, 2004. Available at: so obtained social network. These systems are well known http://www.socsci.auc.dk/~cbruun/aceintro.pdf.
to exhibit interesting characteristics such as robustness, ca- [11] V. Cahill and al. Using trust for secure collabora- pacity of self-adaptation and survivability in uncertain and tion in uncertain environments. IEEE Pervasive Com- dynamic environment and tolerance to randomness. We puting Magazine, special issue Dealing with Uncer- have presented different mechanisms of social behaviours and showed their use in computing environments throughsome illustrative applications. Socially inspired computing [12] Bo Carlsson and Eliasson Gunnar. Industrial dynam- metaphors, provide a real new paradigm for programming ics and endogenous growth. In Industry and Innova- highly distributed and dynamic computing systems. How- tion, pages 1–25. Taylor & Francis, Abingdon, UK, ever, proposed approaches are still developped in an ad hoc manner, and a real theory for socially inspired computing [13] C. Castelfranchi. The theory of social functions: chal- lenges for computational social science and multi-agent learning. Journal of Cognitive Systems Re- [14] Kalyan Chatterjee and Susan H. Xu. Technology dif- [1] Andrew W. Appel and Edward W. Felten. Proof- fusion by learning from neighbours. Advances in Ap- carrying authentication. In ACM Conference on Com- plied Probability, 36(2):355–376, 2004.
puter and Communications Security, pages 52–62,New York, NY, USA, 1999. ACM Press.
[2] O. Babaoglu, H. Meling, and A. Montresor. Anthill: http://www.balancedreading.com/vygotsky.html.
A framework for the development of agent-based ICDCS’02, Vienna, A., July 2002.
In International Conference on Com- [3] Venkatesh Bala and Sanjeev Goyal. Learning from plex Systems (ICCS’04), 2004.
neighbours. Review of Economic Studies, 65(3):595– http://cui.unige.ch/~dimarzo/papers/iccs04.pdf.
[17] M. Dorigo and G. Di Caro. Ants colonies for adaptive routing in packet-switched communication networks.
[4] M. Balze, J. Feigenbaum, and J. Lacy. Decentral- Lecture Notes in Computer Science, page 673, 1998.
ized trust management. In IEEE Symposium on Secu-rity and Privacy, pages 164–173, Los Alamitos, CA, [18] M. Dorigo, V. Maniezzo, and A. Colorni. Ant sys- USA, 1996. IEEE Computer Society Press.
tem: Optimization by a colony of cooperating agents.
IEEE Transactions on Systems, Man and Cybernetics [5] A. Barabasi and R. Albert. Emergence of scaling in random networks. Science, 286(5439):509–512,1999.
[19] M. Faloutsos, P. Faloutsos, and C. Faloutsos. On power-law relationships of the internet topology. In [6] L. Bauer, M. A. Schneider, and E. W. Felten. A Proceedings of the Special Interest Group on Data proof-carrying authorization system. Technical re- Communications (ACM SIGCOMM’99), pages 251– port, Princeton University Computer Science, 2001.
262, New York, NY, USA, 1999. ACM Press.
[20] S. Fenet and S. Hassas. A distributed intrusion de- [31] F. A. Hayek. Studies in philosophy, politics and eco- tection and response system based on mobile au- nomics. Routledge & Kegan, London, 1967.
tonomous agents using social insects communication.
Electronic Notes in Theoretical Computer Science, organization and sorting in collective robotics. Ar-tificial Life, 5(2):173–202, 1999.
[21] S. Fenet and S. Hassas. An ant based system for dy- namic multiple criteria balancing. In Proceedings of [33] Peter Howitt and Robert Clower. The emergence of the Fisrt Workshop on ANT Systems, Brussels, Bel- economic organization. Journal of Economic Behav- ior and Organization, 41(1):55–84, 2000.
[22] Sergio Focardi, Silvano Cincotti, and Michele March- [34] M. Jelasity. Engineering emergence through gossip.
esi. Self-organization and market crashes. Journal In Bruce Edmonds, Nigel Gilbert, Steven Gustafson, of Economic Behavior and Organization, 49(2):241– David Hales, and Natalio Krasnogor, editors, Pro- ceedings of the Joint Symposium on Socially-Inspired [23] N. Foukia, S. Hassas, S. Fenet, and P. Albu- Computing, pages 123–126, Hatfield, UK, 2005. Uni- insect metaphors: a paradigm for distributed intru- [35] M. Jelasity and O. Babaoglu. T-man: Gossip-based sion detection and response systems. In Proceedings of the 5th International Workshop on Mobile Agents G. Di Marzo Serugendo, D. Hales, and F. Zam- for Telecommunication Applications, MATA’03, Mar- bonelli, editors, Engineering Self-Organising Appli- rakech, Morocco, October 2003. Lecture Notes in cations (ESOA’05), Utrecht, The Netherlands, July [24] P. Glansdorff and I. Prigogine. Thermodynamic study of structure, stability and fluctuations. Wiley, 1971.
[36] S. D. Kamvar, M. T. Schlosser, and H. Garcia-Molina.
The eigentrust algorithm for reputation management [25] T. Grandison and M. Sloman. A survey of trust in in p2p networks. In 12th International World Wide internet applications. IEEE Communications Surveys Web Conference : WWW2003, pages 640–651, Bu- [26] P.P Grassé. La reconstruction du nid et les interac- [37] Ulrike Lechner and Beat F. Schmid.
tions inter-individuelles chez les bellicoitermes natal- ties - business models and system architectures: The enis et cubitermes, la théorie de la stigmergie - essai blueprint of mp3.com, napster and gnutella revisited.
d’interprétation des termites constructeurs. Insectes In E. Sprague, editor, the 34th Hawaii International Conference on System Sciences, 2001.
[38] N. Li, J. Feigenbaum, and B. N. Grosof.
Shakhova. Self-organization of trade networks in an logic-based knowledge representation for authoriza- economy with imperfect infrastructure. In Second In- tion with delegation. In 12th IEEE Computer Security ternational Conference on Computing in Economics Foundations Workshop, page 162, 1999.
and Finance, volume 22. Society for ComputationalEconomics, Geneva, Switzerland, 1996.
[39] J.-P. Mano, C. Bourjot, G. Lopardo, and P. Glize. Bio- [28] D. Hales. Choose your tribe! evolution at the next inspired mechanisms for artificial self-organised sys- level in a peer-to-peer network. In S. Brueckner, tems. Informatica, In press, 2006.
G. Di Marzo Serugendo, D. Hales, and F. Zam-bonelli, editors, Engineering Self-Organising Appli- [40] A. Moukas. Amalthaea: Information discovery and cations (ESOA’05), Utrecht, The Netherlands, July filtering using a multiagent evolving ecosystem. Ap- plied Artificial Intelligence, 11(5):437–457, 1997.
[29] Glenn Hardaker and Gary Graham.
[41] S. D. Ramchurn, D. Huynh, and N. Jennings. Trust your e-commerce through self-organising collabora- in multi-agent systems. Knowledge Engineering Re- tive marketing networks. Technical report, School of Business, University of Salford, UK, 2002.
[42] R.Schoonderwoerd, O. Holland, and J.Bruten. Ant- [30] S. Hassas. Using swarm intelligence for dynamic web like agents for load balancing in telecommunications content organization. In Proceedings of the IEEE networks. In Proceedings of the 1st International Swarm Intelligence Symposium, pages 19–25, Los Conference on Autonomous Agents, pages 209–216, Alamitos, CA, USA, 2003. IEEE Computer Society.
ON SELF-ORGANISING MECHANISMS FROM SOCIAL. . .
[43] J. A. Schumpeter. The economy as a whole - seventh chapter of the theory of economic development. In-dustry and Innovation, 9(1/2), 2002.
[44] Jean-Marc Seigneur, Nathan Dimmock, Ciarn Bryce, and Christian Damsgaard Jensen. Combating Spamwith TEA, Trustworthy Email Addresses. In Proceed-ings of the Second Annual Conference on Privacy, Se-curity and Trust (PST’04), pages 47–58, Fredericton,New Brunswick, Canada, October 2004.
[45] Sergey Shumsky. Self-organizing internet semantic network. White paper, NeurOK LLC, 2001.
[46] Max Stewart. The Coevolving Organization. Decom- plexity Associates LtD, Rutland, UK, 2001.
[47] Leigh Tesfatsion. A trade network game with endoge- nous partner selection. In H. Amman, B. Rustem, andA. B. Whinston, editors, Computational Approachesto Economic Problems. Kluwer Academic Publishers,1997.
[48] Leigh Tesfatsion. Agent-based computational eco- nomics: A constructive approach to economic theory.
In K. L. Judd and Leigh Tesfatsion, editors, Hand-book of Computational Economics, Volume 2: Agent-Based Computational Economics, Handbooks in Eco-nomics Series. North-Holland, 2005.
[49] V. A. Vittikh and P. O. Skobelev. Multi-agent sys- tems for modelling of self-organization and cooper-ation processes. In XIII Intern. Conference on theApplication of Artificial Intelligence in Engineering,pages 91–96, Galway, Ireland, 2002.
[50] S. Voulgaris, M. Jelasity, and M. van Steen. A ro- bust and scalable peer-to-peer gossiping protocol. InG. Moro, C. Sartori, and M.P. Singh, editors, Pro-ceedings of Agents and Peer-to-Peer Computing: Sec-ond International Workshop, AP2PC03, volume 2872of Lecture Notes in Artificial Intelligence, pages 47–58, Berlin, 2003. Springer-Verlag.
[51] L.S. Vygotsky. Mind and society: The development of higher mental processes. Harvard University Press,Cambridge, MA, USA, 1978.
[52] S.C. Wang, K.Q. Yan, and C.H. Wei. Mobile target advertising by combining self-organization map anddecision tree. In Proceedings of the IEEE Interna-tional Conference on e-Technology, e-Commerce ande-Service (EEEí04), pages 249–252, 2004.
[53] S. Weeks. Understanding trust management systems.
In IEEE Symposium on Security and Privacy, pages94–105, 2001.
[54] K. Werbach. Syndication: The emerging model for business in the internet era. Harvard Business Review,85:85–93, 2000.

Source: http://inflab.kard.teilar.gr/Informatica_Hassas_etal_final_printed_version.pdf