N. R. Jennings and T. Wittig. ARCHON: Theory and practice. In Distributed Artificial Intelligence: Theory and Praxis, pages 179--195. ECSC, EEC, EAEC, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....agents [59] interface agents [37] virtual agents [1] information agents [35] mobile agents [57] and so on. The diverse range of applications for which agents are being touted include operating systems interfaces [21] processing satellite imaging data [54] electricity distribution management [31], air traffic control [34] business process management [29] electronic commerce [26] and computer games [25] to name a few. Moreover, significant commercial and industrial research and development efforts have been underway for some time [9, 11, 44, 45] and are set to grow further. However, the ....

N. R. Jennings and T. Wittig. ARCHON: Theory and practice. In Distributed Artificial Intelligence: Theory and Praxis, pages 179--195. ECSC, EEC, EAEC, 1992.

....with methodological issues related to the design of human computer interfaces (HCI) for multi agent applications. The focus of the proposed methodological analysis and design framework is on DAI systems made out of coarse grained intelligent agents, like the ones discussed by Jennings and Wittig [18] and Olare and Wooldridge [26] in this volume. These systems consist of complex, discrete expert agents, which have complementary areas of expertise. In general agents within such DAI systems, for example see also [27] 19] 8] adopt coordinated intelligent behaviour patterns, usually ....

....abilities of these systems, related to the diversity of the agents incorporated in them, can provide a wide range of facilities for users. Complex cognitive tasks such as fault detection and diagnosis can be enhanced through the provision of expert information and support as discussed in [18] and [311. The interactive problem solving activity coupled with the high level complexity of such systems, having multiple diverse intelligent agents, with widely distributed knowledge, cooperatively interacting, demands a new approach to interface design. Methodological aspects of this new ....

Jennings N.R. and Wittig T. (1992) *'ARCHON Theory and Practice", in this'volume.

....the functional decomposition into the constituent parts. These systems usually comprise a number of complex heterogeneous coarse gritin agents, which solve through coordinated action a global problem. Typical examples of such systems are the ARCHON application presented by Jennings and Wittig [24] and the FELINE system, by O Hare and Wooldridge [33] in this volume. The user interface of such systems is usually controlled by a dedicated complex agent who schedules dialogue and can hide in various degrees aspects of distribution from the user. This model of interaction is shown in figure 5. ....

Jennings N.R. and Wittig T. (1992) "ARCHON Theory and Practice", in this volume.

....when the underlying genome data is modified by third parties, or when the version of the method employed is updated. 4 Agent Architecture Each agent in the GeneWeaver community shares a common architecture that is inspired by, and draws on, a number of existing agent architectures, such as [21], but in a far more limited and simplified way. An agent contains a number of internal modules together with an external persistent data store that is used for the storage of data it manipulates and external programs used for sequence analysis. In this section, we briefly describe the generic ....

N. R. Jennings and T. Wittig. ARCHON: Theory and practice. In Distributed Artificial Intelligence: Theory and Praxis, pages 179--195. Kluwer Academic Press, 1992.

....DAI and the problem solving methods needed for the application, often compromises are made. Such a compromise very often is to neglect the operating system part by using either easy (and not very efficient) solutions to problems of this level or by employing development (see, for example, ARCHON [JW92]) and testbed (see, for example, DRESUN [CLL93] or MACE [GBH88] systems for DAI applications. Although these systems offer much comfort for implementing various multiagent systems, they have to support often quite different concepts. This results in not very efficient realizations of these ....

Jennings, N.R. ; Wittig, T.: ARCHON: theory and practice, in Avouris, Gasser (eds.): DAI: Theory and Praxis, Kluwer Academic Publishers, 1992, pp. 179195.

....interface agents [10] virtual agents [11] information agents [12] mobile agents [13, 14] and so on. The diverse range of applications for which agents are being touted include operating systems interfaces [15] processing satellite imaging data [16] electricity distribution management [17], air traffic control [18] business process management [19] electronic commerce [20] and computer games [21] to name a few. The richness of the agent metaphor that leads to such different uses of the term is both a strength and a weakness. Its strength lies in the fact that it can be applied in ....

Jennings, N. and Wittig, T. (1992) ARCHON: Theory and practice. In N.M. Avouris and L. Gasser, editors, Distributed Artificial Intelligence: Theory and Praxis, pages 179--195. Kluwer Academic Press, Dordrecht.

....Genome Agent NRDB Agent Protein Calculation Agent Viewer Genome Expert Calc. Agent Fig. 3. GeneWeaver agent community 4 Architecture Each agent in the GeneWeaver community shares a common architecture that is inspired by, and draws on, a number of existing agent architectures, such as [11], but in a far more limited and simplified way. An agent contains a number of internal modules together with either an external persistent data store that is used for the storage of data it manipulates, or the analysis program used to predict function. In this section, we describe the generic ....

N.R. Jennings and T. Wittig. ARCHON: Theory and practice. In Distributed Artificial Intelligence: Theory and Praxis, pages 179--195. ECSC, EEC, EAEC, 1992.

....and languages. Artificial agents are software controlled systems situated in an environment they interact with, and possibly pursuing their goals in a proactive fashion. Agents are being applied in a wide range of domains such as telecommunication networks [2] air traffic control [24] diagnosis [4, 20, 51, 21] and space crafts [50, 34] In this paper, we present a theory, an architecture and several prototype applications of the vivid agent model introduced in [48] 1 City University London, Northampton Square, London EC1V 0HB, Tel. 44 0171 477 8918, msch cs.city.ac.uk 2 Institut fur ....

....executable specifications. Besides the above conceptual requirements which are necessary for adequate agent oriented programming and which motivate the vivid agents framework we will evaluate the theory by practical applications. There has been some work going on to use agents for diagnosis [4, 20, 51, 21, 50, 34, 40, 43]. In the ARCHONproject [4, 20, 51, 21] for example, existing stand alone diagnosis tools have been wrapped up and connected to exchange information and therefore increase the overall performance. In contrast to such top down approaches, NASA develops agents in a bottom up fashion. For the ....

[Article contains additional citation context not shown here]

N. R. Jennings and T. Wittig. ARCHON: Theory and practice. In N. M. Avouris and L. Gasser, editors, Distributed Artificial Intelligence: Theory and Praxis, pages 179--195. Kluwer Academic Press, 1992.

....concerned with the issues of the technology per se but more concerned with the benefits that it delivers, is more likely to sustain the agent paradigm in the longer term. In this respect, work on the development of agent systems for electricity distribution management through the ARCHON project (Jennings and Wittig, 1992), and more recently on business process management in ADEPT (Jennings et al. 1996) for example, provide good illustrations. Similarly, the application of agents to problems in protein analysis and genome structure prediction, in order to make sense of the vast amounts of genomic data that are ....

....of that methodology is likely to be significantly different (Fisher et al. 1997) In either case, however, methodology is an issue that merits much more attention than it has yet achieved. As Crabtree (1998) points out, some initial work resulting from major development projects such as ARCHON (Jennings and Wittig, 1992) has been done, and there are a few more recent efforts in this area, such as the work by Kinny et al. 1996) but these must be just the start of a much more sustained programme. Indeed, as the field matures, the broader acceptance of agent oriented systems will become increasingly tied to the ....

Jennings, NR and Wittig, T, 1992. "ARCHON: Theory and practice". In Distributed Artificial Intelligence: Theory and Praxis, 179--195, ECSC, EEC, EAEC.

....DAI and the problem solving methods needed for the application, often compromises are made. Such a compromise very often is to neglect the operating system part by using either easy (and not very efficient) solutions to problems of this level or by employing development (see, for example, ARCHON [9]) and testbed (see, for example, DRESUN [2] or MACE [7] systems for DAI applications. Although these systems offer much comfort for implementing various multi agent systems, they have to support often quite different concepts. This results in not very efficient realizations of these concepts, ....

Jennings, N.R. ; Wittig, T.: ARCHON: theory and practice, in Avouris, Gasser (eds.): DAI: Theory and Praxis, Kluwer Academic, 1992, pp. 179-195.

....are written by different people, at different times, in different languages, and with different interfaces. Thus, several forms of heterogeneity 1 need to be handled, i.e. different platforms, different program languages, different problem solving paradigm, and the different standards 1 ARCHON [4,5] places considerable emphasis on purpose semantic heterogeneity not deeply considered in the development of our system. The semantic heterogeneity will be one of our future work. used by those services. There have been lots of related approaches to the cooperation of the programs [1,2,3] We ....

N. R. Jennings and T. Wittig. ARCHON: Theory and Practice. In N. M. Avouris & L. Gasser (eds.), Distributed Artificial Intelligence: Theory and Praxis, pages 179-196. Kluwer Academic Publishers, 1992.

....can be applied. This consisted in a collective diagnosis problem among a group of homogeneous robots working at a factory. It should be clear that the scenarios commonly analysed in collective diagnosis research (see for example Frohlich, Nejdl and Schroeder (1998) Bottcher and Dressler (1993) Jennings and Wittig (1992), Schroeder (1998) are somehow different from our example. Our example is much closer to scenarios coming directly from robotics. Nowadays, robots (see McKerrow (1991) for an introduction) regularly substitute humans in many tasks. Diagnosis and maintenance in hazardous environments is one of the ....

Jennings, N. R. and Wittig, T. (1992). ARCHON: Theory and practice, in N. M. Avouris and L. Gasser (eds), Distributed Artificial Intelligence: Theory and Praxis, Kluwer Academic Press, pp. 179--195.

....providing agents. ADEPT is currently being tested on British Telecom (BT) business process which involves some nine departments and 200 different tasks. ffl Industrial systems management. The largest and probably best known European multi agent system development project to date was ARCHON [120, 60, 58]. This ESPRIT funded project developed and deployed multi agent technology in several industrial domains. The most significant of these domains was a power distribution system, which was installed and is currently operational in northern Spain. Agents in ARCHON have two main parts: a domain ....

N. R. Jennings and T. Wittig. ARCHON: Theory and practice. In Distributed Artificial Intelligence: Theory and Praxis, pages 179--195. ECSC, EEC, EAEC, 1992.

.... traffic control [12] and telecommunications network management [76] Indeed, some researchers even go so far as to state that all real systems are distributed [33] Finally, many organisations have substantial amounts of pre existing (or legacy [10] software which could profitably be integrated [42, 44]. This paper discusses the insights gained whilst building a number of industrial multi agent systems under the auspices of the ARCHON project [77] These insights are predominantly based on the experiences from the real world applications of electricity transportation management [43] and ....

N. R. Jennings and T. Wittig, ARCHON: Theory and Practice, in: N. M. Avouris and L. Gasser, eds., Distributed Artificial Intelligence: Theory and Praxis (Kluwer Academic Press, 1992) 179-196.

.... and cooperative problem solving occurs as a consequence of the desire to: increase the level of information integration across organisations [1, 2] overcome the inherent limitations on intelligence present in any finite artificial system [3, 4] develop increasingly complex software systems [5, 6] and provide a more natural representation of the problem being tackled [7, 8, 9, 10] In Distributed Artificial Intelligence (DAI) systems, agents are grouped together to form communities which cooperate to achieve the goals of the individuals and of the system as a whole. Agents are usually ....

N. R. Jennings and T. Wittig, ARCHON: Theory and Practice, in Distributed Artificial Intelligence: Theory and Praxis, eds. N. M. Avouris and L. Gasser (Kluwer Academic Press, 1992) 179-195.

.... this modular approach increases the maintainability of the system, it keeps the overall control at a central location (i.e. a global controller coordinates the activities of all the subcomponents) This centralisation of control has two particular drawbacks for industrial control applications (Jennings Wittig, 1992). Firstly for large applications with a number of distinct supervisory and control subcomponents, the activation of tasks in the sub systems and the decision of what data to exchange between them depends on the state of the entire process. In a centrally controlled system this assessment requires ....

.... acquaintance s skills, interests, current status, Task 1 Task n INTELLIGENT SYSTEM (IS) Monitor Acquaintance Models Self Model High Level Communication Module Planning Coordination Module To and From other agents ARCHON LAYER Figure 1: ARCHON Agent Architecture Task 4 workload and so on (Jennings et al. 1992). These models are essential when coordinating activity because they provide a characterisation of the social problem solving context in which the agent has to operate. Much like the AMs represent other agents in the community, the Self Model (SM) is an abstract characterisation of the agent s ....

[Article contains additional citation context not shown here]

Jennings, N. R. & Wittig, T., (1992). ARCHON: Theory and Practice", In N. M.

....similar benefits see section 4.3 for an account of how this process was carried out for the CERN application. This latter point is important because in many cases developing the entire application afresh would be considered too expensive or too large a change away from proven technology (Jennings and Wittig, 1992). To successfully incorporate both purpose built and pre existing systems, community design must be carried out from two different perspectives simultaneously. A top down approach is needed to look at the overall needs of the application and a bottom up approach is needed to look at the ....

.... one looks after restoration, another tries to diagnose the problem based on different sources of information, and so on and they then communicate relevant information to one another to ensure they are following a coherent course of action towards the overall objective of restoring the service (Jennings and Wittig, 1992). 3.2 Specification of the Agents During normal working conditions, management of the network by the operator in the DCR consists mainly of topology changes (operation on breakers and switches) generation scheduling, and control of 13 the energy interchange with other utilities (Corera et ....

Jennings, N. R., and Wittig, T., (1992) "ARCHON: Theory and Practice" in Distributed Artificial Intelligence: Theory and Praxis (eds. N. M. Avouris and L.Gasser), Kluwer Academic Press, 179-195.

.... ISs can also be incorporated, with a little adaptation, and can experience similar benefits (Jennings et al. 1993) This latter point is important because in many cases developing the entire application afresh would be considered too expensive or too large a change away from proven technology (Jennings and Wittig, 1992). To successfully incorporate both purpose built and preexisting systems, community design must be carried out from two different perspectives simultaneously. A top down approach is needed to look at the overall needs of the application and a bottom up approach is needed to look at the ....

.... one looks after restoration, another tries to diagnose the problem based on different sources of information, and so on and they then communicate relevant information to one another to ensure they are following a coherent course of action towards the overall objective of restoring the service (Jennings and Wittig, 1992). Specification of the Agents During normal working conditions, management of the network by the operator in the DCR consists mainly of topology changes (operation on breakers and switches) generation scheduling, and control of the energy interchange with other utilities (Corera et al. 1993) ....

Jennings, N. R., and Wittig, T., 1992. ARCHON: Theory and Practice. In Distributed Artificial Intelligence: Theory and Praxis (eds. N. M. Avouris and L.Gasser), Kluwer Academic Press, 179-195.

.... the level of information integration across organisations (Pan and Tenenbaum, 1991; Papazoglou et al. 1992; Shina, 1991) overcoming the limitations on intelligence present in any finite artificial system (March and Simon, 1958; Minsky 1985; Simon, 1957) developing sophisticated applications (Jennings and Wittig, 1992; Neches et al. 1991) and providing a more natural representation of distributed problems 1 (eg sensor networks (Lesser and Corkill, 1983) air traffic control (Cammarata et al. 1983) information retrieval (Huhns et al. 1988) and electricity networks (Jennings et al. 1992) Other potential ....

.... applications (Jennings and Wittig, 1992; Neches et al. 1991) and providing a more natural representation of distributed problems 1 (eg sensor networks (Lesser and Corkill, 1983) air traffic control (Cammarata et al. 1983) information retrieval (Huhns et al. 1988) and electricity networks (Jennings et al. 1992)) Other potential advantages include: reusability of problem solving components by incorporating the same system into several cooperating communities, an increased set or scope of achievable tasks by sharing resources, improved system robustness by undertaking duplicate tasks using different ....

[Article contains additional citation context not shown here]

Jennings, N R and Wittig, T, 1992. "ARCHON: Theory and Practice", In N. M. Avouris and L.

.... decomposition and assigning agents to tasks [11] Within the domain of industrial process control, such an approach is infeasible because of the large number of systems which are already in existence and the complexity of the problem being tackled [24] To address this problem the ARCHON project [25], in which some of the work described here took place, focussed on getting possibly preexisting and independent intelligent systems (eg knowledge data bases, numerical systems, etc. to cooperate with each other on a variety of goals. The fact that there is no longer just one aim for the whole ....

....upon objects and react to events in a contingent, immediate, and therefore in the traditional AI sense of the word, unplanned manner. As a result of these insights, a hybrid approach in which both general rules and reactive mechanisms are combined is now being followed within the ARCHON project [25]. GRATE: A General Framework for Cooperative Problem Solving 22 An unexpected spin off of the work in developing generic control and cooperation knowledge, was that a substantial part of a theory for describing how groups of agents should work together emerged. This theory defines the necessary ....

JENNINGS,N.R. & WITTIG,T.: "ARCHON: Theory and Practice", in Distributed Artificial Intelligence: Theory & Praxis, (Ed L.Gasser and N.Avouris), Kluwer Academic Press.

....(1994) for a detailed account of how this process was carried out for the particle accelerator control application. This latter point is important because in many cases developing the entire application afresh would be considered too expensive or too large a change away from proven technology (Jennings and Wittig, 1992). To successfully incorporate both purpose built and pre existing systems, community design must be carried out from two different perspectives simultaneously. A top down approach is needed to look at the overall needs of the application and a bottom up approach is needed to look at the ....

Jennings, N. R., and Wittig, T., (1992) "ARCHON: Theory and Practice" in Distributed Artificial Intelligence: Theory and Praxis (eds. N. M. Avouris and L.Gasser), Kluwer Academic Press 179-195.

....However pre existing ISs can also be incorporated, with a little adaptation, and can experience similar benefits. This latter point is important because in many cases developing the entire application afresh would be considered too expensive or too large a change away from proven technology (Jennings and Wittig, 1992). To successfully incorporate both purpose built and pre existing systems, community design must be carried out from two different perspectives simultaneously (Varga et al. 1994) A top down approach is needed to look at the overall needs of the application and a bottom up approach is needed to ....

....who would benefit from receiving it. The PCM is composed of generic rules about cooperation and situation assessment which are applicable in all industrial applications all the domain specific information needed to define individual behaviour is stored in the self and acquaintance models (Jennings, 1992). The former contains information about the local IS and the latter contains information about the other agents in the system. The type of information contained in both models is approximately the same, although it varies in the level of detail, and includes the agent s skills, interests, current ....

Jennings, N. R., and Wittig, T., (1992) "ARCHON: Theory and Practice" in Distributed Artificial Intelligence: Theory and Praxis (eds. N. M. Avouris and L.Gasser), Kluwer Academic Press 179-195.

.... and virtually reality systems (Bates, 1994) characters can naturally be represented as self contained, autonomous, social problem solving entities (i.e. agents) Legacy Systems In many traditional areas of computing (such as computer integrated manufacturing (Parunak, 1995) and process control (Jennings and Wittig, 1992)) there is a significant Applying Agent Technology 8 amount of existing software (especially information systems) which performs critical organisational functions. To keep pace with changing business needs, this software must be periodically updated. However, modifying such legacy systems is ....

N. R. Jennings, and T. Wittig. 1992. ARCHON: theory and practice. In Distributed Artificial Intelligence: Theory and Praxis (eds. N. M. Avouris and L. Gasser), Kluwer Academic Press pp 179-195.

.... and virtually reality systems (Bates, 1994) characters can naturally be represented as self contained, autonomous, social problem solving entities (i.e. agents) Legacy Systems In many traditional areas of computing (such as computer integrated manufacturing (Parunak, 1995) and process control (Jennings and Wittig, 1992)) there is a significant amount of existing software (especially information systems) which performs critical organisational functions. To keep pace with changing business needs, this software must be periodically updated. However, modifying such legacy systems is in general very difficult, since ....

N. R. Jennings, and T. Wittig. 1992. ARCHON: theory and practice. In Distributed Artificial Intelligence: Theory and Praxis (eds. N. M. Avouris and L. Gasser), Kluwer Academic Press pp 179-195.

.... an increasingly large number of applications are being conceptualised in terms of cooperating agents examples include speech processing (Erman Lesser, 1975) flexible manufacturing systems (Parunak, 1987) air traffic control (Cammarata et al. 1983) electricity transportation management (Jennings et al. 1992), telecommunications network management (Weihmayer Brandau, 1990) design (Klein, 1991) concurrent engineering (Reddy et al. 1993) sensor interpretation (Lesser and Erman, 1980) and particle accelerator control (Jennings et al. 1993) In such Distributed Artificial Intelligence (DAI) systems ....

.... domain level problems (e.g. fault diagnosis, security analysis and restoration planning) and another for participating in environments containing other entities (e.g. being able to request information and services from acquaintances and sending timely information to help community members) Jennings and Wittig (1992) highlight two features of industrial applications which have a significant impact upon the design of a cooperating community. Firstly there is a vast amount of pre existing software currently being used. Secondly a given application is composed of a diverse range of generic tasks (e.g. diagnosis, ....

[Article contains additional citation context not shown here]

Jennings, N. R. & Wittig, T., (1992). ARCHON: Theory and Practice", In N. M. Avouris & L.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC