J Intell Manuf (2007) 18:529542 DOI 10.

1007/s10845-007-0063-3

A comprehensive modeling framework for collaborative networked organizations

Luis M. Camarinha-Matos Hamideh Afsarmanesh

Published online: July 2007 Springer Science+Business Media, LLC 2007

Abstract Collaborative networked organizations (CNOs) are complex entities whose proper understanding, design, implementation, and management require the integration of different modeling perspectives. A large number of modeling tools and theories that have been developed in other disciplines have a potential applicability in this domain. Therefore, an identication of the most promising approaches is made and mapped into four dimensions of an endogenous perspective of collaborative networked organizations: structural, componential, functional, and behavioral. But a comprehensive modeling of such complex dynamic systems requires also an exogenous perspective, the life cycle dimension, and a stratication of models according to the modeling purpose. Thus a comprehensive modeling framework is therefore proposed as a rst step towards the elaboration of a reference model for collaborative networks. Keywords Collaborative networks Virtual organizations Virtual enterprises Modeling framework

Introduction Modeling is one of the key activities in understanding, designing, implementing, and operating systems. Modeling is at the very heart of any scientic and engineering activity. When a team of researchers or system designers develop a new system, the output of the design phase is a model or set of
L. M. Camarinha-Matos (B) New University of Lisbon, Quinta da Torre, 2829-516 Monte Caparica, Portugal e-mail: cam@uninova.pt H. Afsarmanesh University of Amsterdam, Amsterdam, The Netherlands e-mail: hamideh@science.uva.nl

models of the system to be implemented. A model, as an abstract representation of the intended system, will then be used to guide the implementation. Due to a number of practical contingencies, the implemented system might show some (minor) differences regarding the original model (usually the case). Furthermore, a model is also very useful in order to supervise (manage) the operation of the developed system during its life cycle. Complementarily, a model can also be used to predict the behavior of the system being developed or managed. As in any other scientic discipline or engineering branch, collaborative networked organizations (CNOs) require the development of models, not only as a help to better understand the area, but also as the basis for the development of methods and tools for better decision-making. In fact proper decision-making in all phases of the CNO life cycle needs to be based on well argued and veried models and methodologies. These models and methodologies constitute the basis for the ICT-based support for business and organizational development and operation, as well as the base for education, training, and effective operation of CNOs. CNOs are complex systems, emerging in many forms in different application domains, and consist of many facets whose proper understanding requires the contribution from multiple disciplines. However, an analysis of past modeling efforts indicates that practitioners and researchers are not fully aware of a comprehensive spectrum of suitable modeling processes, tools, and methodologies. For instance, very often modeling is restricted to a processes view (e.g. SCOR type of models for traditional supply chains). Or they stick with one approach such as using UML even though it might not the most appropriate approach for all or a part of the modeling effort. This situation is however improving and lately some theories and paradigms dened elsewhere have been suggested

123

530

J Intell Manuf (2007) 18:529542

by several research groups as promising tools to help understand and characterize emerging collaborative organizational forms (Camarinha-Matos & Abreu, 2003; Camarinha-Matos & Afsarmanesh, 2004; Eschenbaecher & Ellmann, 2003). Nevertheless, it is unlikely that any of these theories and modeling methods will cover all needs of CNO; they can be used as a starting point but extensions or adaptations are needed. There is no single formal modeling tool/approach that adequately covers all perspectivesno universal language for all problems. And yet existing knowledge on diverse manifestations of traditional collaborative networks is quite fragmented, being urgent to proceed with an integration and formalization effort. Nevertheless, purely formal methods in addition to being hard to apply are also difcult to follow by those not familiar with such methods. Dissemination and communication is one important purpose for modeling CNOs. As such, we must acknowledge that this area is addressed by a large variety of people with different backgrounds; not all of them possess a strong formal background, and even many of the ICT practitioners do not have a formal education on computer engineering or computer science. This might suggest, in some cases, the appropriateness of semi-formal methods. On the other hand, new forms of collaborative networks and new patterns of behavior are being invented and explored, for which it is not feasible to develop fully consistent formal models at start. In these cases, semi-formal models, or even informal analogies as represented by metaphors, can provide valuable insights towards a preliminary level of understanding of new collaborative forms. This paper analyses a set of relevant modeling needs for CNOs, extending previous partial contributions (CamarinhaMatos & Afsarmanesh, 2006a, b) considers a collection of tools and theories developed elsewhere that might be useful here, and suggests a modeling framework for CNOs integrating multiple perspectives.

of common operating principles and infrastructures, with the main goal of increasing their preparedness towards collaboration in potential Virtual Organizations (Afsarmanesh & Camarinha-Matos, 2005). Certainly many other relevant questions may be asked in relation to a VBE. Similarly, for VO management a large number of modeling purposes are typically considered (bottom part of Fig. 1). In the same way many purposes are identiable for Professional Virtual Communities (PVCs) and other forms of CNOs. Given this large diversity of modeling purposes, which also leads to different types of models, it is important to establish a framework for modeling that structures and guides the modeling process.

CNO modeling dimensions As a rst attempt to reach a comprehensive modeling framework for CNO modeling, four dimensions are proposed as follows:

Modeling needs in CNO In the context of a complex system like a CNO, modeling is fundamental for understanding, managing, simulating/predicting the behavior of CNOs, and certainly also for software development. For instance, in the VOSTER project (Lh, Zhang, & Katzy 2005) several purposes for modeling in this domain were also considered. In ECOLEAD a large number, though incomplete, of modeling purposes were identied for various kinds of CNOs. Based on those examples, the top part of Fig. 1 illustrates some of the important questions a modeler may pose when attempting to model a virtual organizations breeding environment (VBE). A VBE is an association of organizations and their related supporting institutions, adhering to a base long term cooperation agreement, and adoption

Structural dimension. This perspective addresses the structure or composition of the CNO in terms of its constituting elements (participants and their relationships) as well as the roles performed by those elements and other characteristics of the network nodes such as the location, time, etc. This perspective is used in many disciplines (e.g. systems engineering, software engineering, economy, politics, cognitive sciences, manufacturing), although with different wording and diversied tools. Componential dimension. This dimension focuses on the individual tangible/intangible elements in the CNOs network, e.g. the resource composition such as human elements, software and hardware resources, information and knowledge. Not all these elements are physical in a strict sense but rather represent the things of which the network is built of. Furthermore, the componential dimension also consists of ontology and the description of the information/knowledge repositories that pertain to the CNO. Functional dimension. This perspective addresses the base operations available at the network and the execution of time-sequenced ows of operations (processes and procedures) related to the operational phase of the CNOs life cycle. Behavioral dimension. This dimension addresses the principles, policies, and governance rules that drive or constrain the behavior of the CNO and its members over time. Included here are elements such as principles of collaboration and rules of conduct, contracts, conict resolution policies, etc.

123

J Intell Manuf (2007) 18:529542 Fig. 1 Examples of modeling purposes

How do I create my VBE?
-Members selection process? -Governance rules and procedures? -Responsibilities & liabilities model? -Contracts and agreements model?

531

VBE Modeling Purposes ? Constructs & Tools ?

How do I manage my VBE?

-Members directory
- Profiles, competencies, roles, ...?

-Performance histroy? -Catalog of products / services? -Indicators performance, readiness, ...? -Value system model? -Trust model? -VBE marketing elements?

Roles
How do I evolve / metamorphose my VBE?

How do I create a VO?

-Business opportunity model? -Profiles of potential partners? -Performance history of partners? -Contract models / templates? -Negotiation process? -Constraints? IPR? Benefits model?

Structures Directories Rules Principles Functions Processes

Constraints

VO Management Modeling Purposes ? Constructs & Tools ?

Templates

Taxonomies Ontologies

What about my VO?

-Who are the partners & roles? -What is the business goal? -Business processes model? -Competencies & resources? -Contract & agreements model?

What about performance?

-Records & indicators? -Events & exceptions? -Supervision process? -Quality control process?

Resources

How do I manage my VO?

-Process models? -Decision support mechanisms? -Conflict resolution process? -Optimization methods? -Risk management procedures? -Inheritance principles? -Performance records? -Property rights model? -Liabilities model?

How do I dissolve my VO?

These specic dimensions are chosen for the reason of their "near-orthogonality" in the sense that they are primarily disjoint in dividing this sub-space, and that if elements in different dimensions are bound to each other, then changes in one dimension affect the elements of the other dimensions weakly across some region of relevance. For example, extending the number of workers in one organization (a componential element of modeling an organization) may change the options in a process workow (a functional element of modeling that organization). As such, with these four dimensions every CNO can be comprehensively dened (modeled) by the collection of its four models, as well as a set of (weak) bindings across the constituents of those models. Every model represents specic (and orthogonal) aspects/perspective/dimension of a CNO. Two examples for bindings follow: 1Dependencies and bounds between the componential components (e.g. the personnel) and the structural model counterpart (e.g. the role and skill of the personnel) within a CNO. 2Connection between an organizations structural component (e.g. rights/ duties of the organization in a VO) and the behavioral model counterpart (e.g. the contract component in the VO). The suggested dimensions are still very general and it is important to consider a ner level of granularity; in other words, to consider a set of sub-dimensions for each dimension. Therefore the following set of sub-dimensions is initially proposed for a CNO modeling framework: 1. Structural dimension (a) Actors/relationshipsidentifying all the participating actors (nodes) in the network as well as their inter-relationships (arcs). The actors can be enterprises, other types of organizations, or people. Two

(or more) actors can be linked through a number of different types of relationships, e.g. client-supplier, sharing, co-authoring, etc. (b) Rolesdescribing and characterizing the roles that can be performed by the actors in the network. A role denes an expected behavior for an actor in a given context. Examples of roles include member, coordinator, broker, planner, etc. 2. Componential dimension (a) Hardware/software resourcescharacterizing the equipment, software, and infrastructures used/shared in the network. In terms of ICT equipment this model can include the architecture of the computer network supporting the collaboration. In the case of manufacturing networks it can include the layout of the shared facilities as well as the logistics networks. (b) Human resourcesa characterization of the human resources available in the network, namely in terms of their competencies, prole, potential roles they can perform, etc. (c) Information/knowledge resourcesunder this subdimension we can include the repositories of information and knowledge that are shared by the network members or that support the collaboration processes and the networked organization. (d) Ontology resourcesrepresenting the main (common) ontologies used in the network and that facilitate the mutual understanding among the network members. One example can be the ontology of competencies available in the network. 3. Functional dimension (a) Processesthis sub-dimension is concerned with the processes involved in the main line of activities of

123

532

J Intell Manuf (2007) 18:529542

the collaboration. Processes represent the main structured part of the operational activities of the network. An example is the distributed business processes in a business oriented CNO. (b) Auxiliary processesincluding those processes that are designed to assist the CNO in terms of its maintenance and improvement of operations. Examples include performance monitoring processes, competencies management processes, etc. (c) Methodologiestypically less formalized than processes, represent the body of practices, procedures, and rules used mainly by human actors in a CNO. They are frequently represented as a semi-structure set of steps (informal enumeration of activities) combined with some structured representation of input/ output information. An example can be the methodology to be followed by a broker to announce a business opportunity to the CNO members. 4. Behavioral dimension (a) Prescriptive behaviorcapturing the elements that lie down or prescribe normative guidelines or rules for the proper behavior of the CNO such as (general) principles, strategies, and protocols. An example is a recommendation for CNO members to give preference to network peers when searching for partners for a business opportunity. Another example could be the recommended protocol when negotiating a contract. (b) Obligatory behaviordescribing those rules and principles that are mandatory to be followed inside the network. This includes policies, governance values and associated rules, and enforcement steps. An example can be the internal rules used for distribution of benets or for sharing the operational costs of the network. (c) Constraints and conditionsrepresenting those environmental features that limit the context of operation of the CNO and its members. An example is a set of restrictions on the use of intellectual property of one member by other members of the network. (d) Contracts and cooperation agreementscovering both the contracts between the CNO and external customers and the internal contracts and cooperation agreements among the network members. These models may include both representations understandable to humans and to software systems.

They are shown mainly to better characterize, by illustration, the scope of each dimension.

Map of potential approaches A large number of theories and tools, developed elsewhere, are potentially useful in modeling CNOs or in giving insights to better understand these networks. It would be a matter of practical convenience for the CNO community to have a kind of map or shopping list relating such tools and theories to the CNO modeling needs or modeling dimensions. In this direction, Table 4 illustrates the potential applicability of the various theories with respect to the four modeling dimensions introduced above. In this table the letters [SD], [CD], [FD], [BD] stand for Structural, Componential, Functional, and Behavioral dimension, respectively. In addition to these theories and associated tools, there are other modeling tools that have a generic applicability or have been already widely used in the CNOs community and therefore were not studied in this work. Nevertheless they shall be considered as important candidates to some or all of the modeling dimensions. These generic tools include (Table 5) Figure 2 represents a simplied attempt to establish a map relating theories/tools to the modeling dimensions. This map is not exhaustive and certainly not fully accurate, but just a contribution to give a rough idea of the many possibilities that can be considered. Some theories and tools have a generic nature, others are very specic. For instance, UML or formal engineering methods are quite general and thus potentially applicable to all modeling dimensions; on the other hand, deontic logic is quite specic and potentially useful in the modeling of some aspects of behavior. As also shown in Fig. 2, there are some sub-dimensions for which there is no specialized theory that is particularly suited (e.g. hard/soft resources). Nevertheless there are some generic theories/tools (associated to CNO modeling framework in the center) that are "good for everything", like UML, ontology, etc. Another aspect to consider is that some theories might cover, in part, more than one dimension or subdimension. For instance, complexity theories can be linked to the functional and behavioral dimensions. Not all these possibilities are represented in Fig. 2. The suitability of a theory/tool to be applied to a particular modeling perspective also depends on the experience of the modeler with that theory/tool. There are in fact several gray areas of applicability. For instance, self-organizing systems could, in a limited way, also relate to the structure of the network. Therefore, and in order to not make the map too complex, only what currently seems to be the most important and obvious links are represented.

Tables 13 illustrate these dimensions and sub-dimensions in the case of a VO breeding environment, VO management, and professional virtual communities (PVC). Nevertheless it shall be noted that these sub-dimensions are not exhaustive.

123

J Intell Manuf (2007) 18:529542 Table 1 Application example for a VO breeding environment Dimension Structural Sub-dimension Actors/relationships Roles Componential Hardware resources Software resources Human resources Info /Knowledge Ontology resources Functional Processes Examples

533

VBE members (business entities, non-prot institutions, support institutions), collaboration links, trust links, coordination links, etc. VBE member, VBE administrator, Broker, VO planner, VO coordinator, VBE advisor, etc. (Domain specic) machines, ICT infrastructure Common software tools, VBE specic services, commercial tools with special agreement with VBE, etc. Human actors performing a role in the VBE. Directory of members, Record of past performance, Directory of running VOs, Ranking and Ratings, Patents, etc. Common ontology, Prole and competencies ontology, etc. VBE creation processes (initiation & recruiting, foundation), Trust management process, Competency management process, Marketing & promotion process, VO creation process, Metamorphosis process, etc. Assets management process, VBE Monitoring process, Incentive creation process, Competencies assessment process, etc. Conict resolution procedure, Performance management procedure, Members registration procedure, Competences and Proling management approach, etc. Membership eligibility principles, Brokering principles, Decision-making principles, Rewarding and Sanctioning principles, Leadership principles, Commitment rules, etc Bylaws, VBE behavior rules (Ethical code and culture), Governance policies, etc. Constraints imposed by the external environment, Constraints on the use of some resources, Constraints on timing for interactions (when partners in different time zones), Legal constrains to VBE due to differences in government laws among countries, possible cultural constrains, etc. VBE adhesion contract, VBE cooperation agreement, Contract templates for VOs, etc.

Aux. processes Methodologies Behavioral Prescriptive behavior Obligatory behavior Constraints & conditions

Contracts & agreements

Table 2 Application example for VO management Dimension Structural Componential Sub-dimension Actors/relationships Roles Hardware resources Software resources Human resources Info/Knowledge Ontology resources Functional Processes Aux. processes Methodologies Behavioral Prescriptive behavior Obligatory behavior Constraints & conditions Contracts & agreements Examples VO members, VO topology, Control ow and Material ow links, etc. VO broker, VO planner, VO manager/coordinator, VO member, etc. VO support (shared) machinery, ICT infrastructure, etc. VO management services, common collaboration support tools, etc. Human actors performing a role in the VO Product/service model, Business process model, Performance data, lessons learned, etc. General (common) ontology, Prole and competencies ontology of VO members, etc. Planning processes, Fulllment and alerting process, Performance measurement process, VO evolution process, VO dissolution process, etc. VO contract management process, VO re-organization process, Supporting VO transparency process, Simulation processes, etc. Conict resolution procedure, Decision making procedure, Decisions communication procedure, etc. VO management approach, Prot distribution rules, Risk sharing rules, etc. Activity reporting rules, Performance reporting rules, Contractual constraints, Resource use constraints, etc. VO contract, VO consortium agreement, etc.

123

534 Table 3 Application example for PVC Dimension Structural Sub-dimension Actors/relationships Roles Componential Hardware resources Software resources Human Resources Info/Knowledge Ontology resources Functional Processes Examples

J Intell Manuf (2007) 18:529542

PVC Member (knowledge workers, freelancers, employers of a VBE enterprise)/Social collaboration, sharing knowledge, looking for business opportunities, etc. PVC Initiator, Broker, PVC Member, PVC Administrator, etc. ICT infrastructure. Advanced Collaborative PlatformPVC services, tools, etc. Human resources (professionals and auxiliary staff). Members prole (skills & competencies) info, VT and member record of past performance, members IPR info, etc. PVC general body of knowledge ontology, Member ontology (prole), Business ontology, Tree of competencies/skills ontology (member & PVC), etc. PVC start up process, Members subscription process, Trust management process (socializing process), Competency management process (knowledge process), Marketing and promotion process, VT creation and metamorphosis processes (business process), etc. Performance indicators management process, Ontology management process, PVC Administrative process, Members internal evaluation process (competences, successes, etc), Motivation induction process, Inheritance process, etc. Subscription procedure, Steps for initializing a PVC, Business announcement procedures, Business negotiation procedures, Conict resolution procedure, etc. Social & knowledge sharing protocols (way of conduct), Business negotiation protocols, etc. The PVC Bylaws, PVC governance policies, etc. IPR rules, Penalties corresponding to the PVC rules breach, environment constraints (access rights), etc. Membership agreement, Templates for the VT agreement, Templates for the Internal project contract (PVC adhesion contract), Templates for the PVC external contract, etc.

Aux. processes

Methodologies Behavioral Prescriptive behavior Obligatory behavior Constraints & conditions Contracts & agreements

Another relevant observation is that some theories also propose a specic modeling language/tool (e.g. the theory of Petri nets suggests a specic graphical modeling language); others dont and leave it up to the modeler to chose a representation formalism. The level of maturity of each approach/ theory is also different (Table 6). On the other hand not all theories and modeling approaches are at the same level. In fact some of them are specializations of others or show a strong dependency on others, as illustrated in Fig. 3. As the mentioned theories and modeling approaches were originated in different scientic disciplines, it is not easy to fully inter-relate them. Nevertheless this diagram can help the reader in locating each theory or modeling approach, i.e. getting a rough idea of what it covers, through association with others. For instance, a reader not familiar with Social Network Analysis can see that this theory is a kind of specialized branch of Graph Theory. A recurrent debate can be found on the appropriate level of formalization to be used in the models. In other words, what is the desired level of formalization to be applied in CNOs? From a purely theoretical point of view, it is well known that formal models, due to their strong mathematical foundation, are likely to reduce ambiguity and errors and thus are generally preferable as a basis in consolidated disciplines. However, the actual answer depends on the background of the

modeler, the purpose and scope of the model, among other factors such as the level of maturity of the knowledge in the considered domain. Formal methods facilitate automatic verication of the properties of the system being modeled as well as automatic synthesis of system components. But they are quite hard to handle by humans, even those with and engineering background. Informal and semi-formal methods not always give adequate results in terms of correctness and precision. On one hand, informal methods normally use natural languages for which are easier to understand by human interlocutors; they are however problematic when condence in correctness is desired. On the other hand, semi-formal graphical representations such as data ow diagrams are easily readable, and provide some sort of organization or structure to the descriptions, but lack precise semantics, which may lead to ambiguous interpretations and turn reasoning or automation processes more difcult. Attempts to introduce formal methods have been exercised in the areas of systems automation and software engineering, namely for automatic synthesis and verication. Many researchers in these areas claim that the more formal a model is the more correct, complete, precise, and unambiguous the resulting system remains. Furthermore, some state that

123

J Intell Manuf (2007) 18:529542 Table 4 Some theories and their potential applicability in CNOs Theory/Tool Benchmarking Complexity theories Potential contribution to CNO modeling

535

[FD] Assessment of performance in comparison with a reference (benchmark), including assessment of processes, trustworthiness, and suggestion of best practices. [FD] Methods for forecasting emergent behavior, trustworthiness, etc. [BD] Modeling of emergent behavior in advanced networks. Qualitative (macro) understanding of CNOs life cycle.

Decision support Deontic logic Distributed group dynamics Diversity in work teams Evolving ontologies Federated systems

[FD] Give a basis for developing methods to assist humans in decision making. [BD] Represent in a formal way aspects such as it is obligatory that , it is forbidden that , it is permitted that , which can be useful in the governance of behavior. [SD] Focus on inter-group relationships such as power, leadership, etc, [BD] Analysis of leadership behavior, hostility, compliancy, etc. [SD] Characterization of the diversity of individuals and cultures found in CNOs and analysis of the potential induced by this diversity. [CD] To capture the evolution of mutual understanding among members of the network, but still is offering limited results. [SD] Providing a vision of the CNO as a federation of autonomous, heterogeneous, and distributed sources of resources (data /information, services). Relate roles with authorized access to and visibility of resources. [CD] Distributed data/information repositories. [SD] [CD] [FD] [BD] Rigorous specications (mathematical-based) with potential application in verication and synthesis of systems. Very hard to apply. [SD] [CD] [FD] [BD] Solve design problems (architecture, protocols, verication of specications according to correctness and completeness), but very hard to develop. If developed for specic perspectives/subsystems, can contribute to reduce ambiguities and provide a sound basis for further developments. [FD] Can provide concepts for decision-making, e.g.: Cooperative game theory: distribution of responsibility and resources. Non-cooperative game theory: selection of partners, sustaining cooperation and trust building. [BD] Model interactions with formalized incentive structures. [SD] Representation of the structure of the network topology, routing, activity, ow. [FD] Methods to perform computations on ows and optimization. [ CD] Providing visual representations of knowledge which can facilitate analysis of the CNO and its resources. [ BD] Help understanding some aspects of the dynamics of evolutionary processes (cognitive and business) in multi-cultural contexts. [SD] [CD] [FD] [BD] Quick description for human communication namely a possible help in expressing complex ill-dened concepts. Can be used in early stages (conceptual design) as long as they are not taken too literally. [FD] Use of probabilistic inference to update and revise belief values. Can support complex inference modeling including rational decision making systems, value of information and sensitivity analysis. Causality analysis and support a form of automated learning (parametric discovery, network discovery, and causal relationships discovery).

Formal engineering methods Formal theories

Game theory

Graph theory Knowledge mapping Memetics Metaphors

ML/Bayesian networks

Multi-agent systems

[FD] [BD] Model societies of autonomous, distributed and heterogeneous entities, giving insights on how these societies can be organized and their behavior regulated through norms and institutions. [FD] Brokering, coalition formation and negotiation. [BD] Simulation of self-organizing behavior.

Multi-agent dependency theory Network analysis Portfolio theory Real options theory Scopos theory

[FD] [SD] Representation of social interactions among agentsdependency relations, power relations. [SD] [FD] Specialized graph theory-based algorithms for application in network management systems (mostly applied in telecommunication networks). [FD] Decision making such as in VO creation (to select the optimal VO from a VBE) [FD] Decision making, e.g. decision to create a VO for a business opportunity, evaluation of the minimum protable bid in a call for tenders, etc. [FD] Understand transformation of information or knowledge from one cultural and language environment to others in such a way that the understanding and conception of the source information or knowledge would be the same for all.

123

536 Table 4 continued Theory/Tool Self-organizing systems Semiotics Social network analysis Soft computing Synergetics Temporal and modal logic Transactions cost theory Trust building models Web & text mining Potential contribution to CNO modeling [BD] Understanding and simulation of self-organizing behavior. [FD] Help in predicting evolution. [BD] Model responsibility relationships and commitments. Prescribe norms and rolesepistemic, deontic and axiologic.

J Intell Manuf (2007) 18:529542

[SD] Analysis of social and organizational structure of CNOs, including provision of a number of metrics. Ongoing research may lead to useful results on the inclusion of soft-modeling aspects. [FD] [BD] Represent and exploit the tolerance for imprecision, uncertainty, partial truth, and approximation. Particularly important to model human and social aspects. [BD] Help understanding emerging behavior and emerging values. [FD] [BD] Focus on the representation of temporal information within a logical framework. Can be used to model temporal aspects of processes and some aspects of behavior. [FD] Understand and analyze governance structures based on transaction costs. [FD] Organize and systematize the trust building and trust management processes. [FD] Analysis and knowledge discovery from unstructured data: documents in free text form, web documents. Potential applications include evolution of ontologies, nding business opportunities, etc.

Table 5 Additional tools and their potential applicability in CNOs Ontology Petri nets Workow UML [SD] [CD] [FD] [BD] Representation of the main CNO concepts and their relationships. [FD] Modeling or processes and auxiliary processes. [FD] Modeling or processes and auxiliary processes. [SD] [CD] [FD] [BD] Generic object-oriented modeling tool (graphical language) with potential application to all dimensions of CNO. However, being a generic tool, it does not properly capture all specicities of each dimension.

Fig. 2 An attempt to map modeling theories applicable to CNOs

Social network analysis Distrib. group dynamics Federated Systems

Graph theory

Multi Agent dependency

Network Analysis Formal theories Formal Enginr methods UML

Hard / soft resources Human resources

Federated Systems

Actors & relationships

Structural dimension
Roles

Componential dimension

Info/Knowledge resources Ontology resources

Knowledge mapping

Diversity in work teams Workflow Scopos theory Petri nets

Processes Auxiliary processes

Evolving ontologies Federated Systems Game theory Semiotics

CNO Modeling Framework

Ontology

Network Bench Analysis marking Decision support Game theory ML / Bayesian Portfolio theory net

Prescriptive behavior

Functional dimension Metaphors

Behavioral dimension

Obligatory behavior Constraints & conditions Contracts & agreements

Memetics

Methodologies

Graph theory Real options theory

Distributed Group dynamics Deontic logic

Multi Agent dependency Multi Agent systems

Complexity theories Temporal Soft & modal computing logic Syner getics

Transaction Web & Soft Cost Text Trust computing theory mining build models

Self organizing systems

Modeling dimension Taxonomy of modeling dimensions Suitable to model the dimension Modeling sub-dimension Modeling tool / theory

123

J Intell Manuf (2007) 18:529542 Table 6 Some characteristics of considered theories

Theory / Tool Benchmarking Complexity theories Decision support Level of formalization Includes specific modeling language / Examples (Statistical methods) Decision Trees Multi-attribute Modeling Bayesian networks Logic-based Maturity level

537

Deontic logic Distributed group dynamics Diversity in work teams Evolving ontologies Federated systems Formal engineering methods Formal theories Game theory Graph theory Knowledge mapping Memetics Metaphors ML/ Bayesian networks Multi-agent systems Multi-agent dependency theory Network analysis Portfolio theory Real options theory Scopos theory Self-organizing systems Semiotics Social network analysis Soft computing Synergetics Temporal and modal logic Transactions cost theory Trust building models Web & text mining Other tools Ontology

Z, VDM, RAISE, Estelle, First order languages, set theory (Mathematical) Mathematical / set theory, graphical (Diversity of graphical tools)

Probabilities (FIPA standards), A-UML

(Mathematical / Probabilities)

(Graph theory + statistics) Fuzzy Logic, Probabilistic Reasoning, Neural Computing, Genetic Computing Logic-based

Text mining methodologies and techniques; Text-mining applications (e.g. TOKO) Ontology engineering methodologies and techniques; Ontology representation frameworks (e.g. OWL, RDFS); Ontology editors (e.g. Protg, Triple20) Graphical, algebraic Graphical, (Visio, Rat. Rose) XPDL, WS-BPEL,

Petri nets UML Workflow

Level of formalization: Semi-formal Formal Informal / descriptive Specific modeling language: No (resorts to a variety of languages) Yes (Which? examples) Maturity level: On-going research Mature &still being extended Stable and mature

formal methods with properly dened semantics can predict potential solutions to the systems behavior resulting in a more efcient and easily maintainable system (Alexander, 1996; Bowen & Hinchey, 2006; George & Vaughn, 2003; Liu, 2005). Nevertheless, difculties to employ formal methods emerge when one tries to formalize a complex system as a whole. According to (Bowen & Hinchey, 2006), when analyzing the state practice of adoption of formal models in the last 10 years, many highly publicized projects proclaimed as great formal methods successes formalized only 10 percent or less of the system.

Since a complete formalization of the properties of a whole (complex) system is not feasible in practice, it would be interesting to think of combining the best of traditional models with the best of formal methods in order to achieve a semirigid level of formalization for CNOs. In fact the modeling scope in CNOs does not only focus on the software components but rather involves many other perspectives of a socioorganizational nature that are hard to model. There is also a large variety of purposes and users for the developed models. Therefore, a realistic approach is to consider that most modeling efforts at this stage have to be based on semi-formal

123

538

J Intell Manuf (2007) 18:529542

use

Formal Theories Graph Theory Emergent Behavior

Formal (Engineering) Methods Network Analysis

Probabilistic modeling

Logic

Social Network Analysis Complexity Theories SelfOrganizing Systems

Bayesian Networks

Soft Computing

Temporal & Modal Logic

Deontic Logic

Distributed & Autonomous systems Federated Systems Multi-agent Systems Multi-agent Depedency Theory Ontology

Knowledge & Info modeling

Activity / Data Flow languages

can use

Knowledge Mapping

Semiotics

UML

Petri Nets

Workflow Systems

Evolving Ontology

Web & Text Mining Qualitative methods Metaphors Distributed Group Dynamics Transactions Cost Theory Synergetics Diversity in Work Teams Scopos Theory

Decision Support Methods

Benchmarking Game Theory

Portfolio Theory

Real Options Theory

Memetics

Fig. 3 Modeling islandssome dependency relationships between theories

approaches. To a lesser extend, and in specic sub-areas, formal models can be progressively attempted in order to facilitate both the scientic progress and the engineering activities. Towards a comprehensive modeling framework When modeling a CNO, it is important to consider both its internal and external aspects (Fig. 4) i.e. how to see the network from inside (as in traditional systems modeling) and from outside (i.e. the interactions between the CNO and its surrounding environment) (Camarinha-Matos & Afsarmanesh, 2006a, b). We can therefore consider two modeling perspectives: endogenous elements and exogenous interactions: Endogenous elements This perspective aims at providing an abstract representation of the CNO from inside, namely the identication of a set of characteristic properties that can together capture the elements constituting CNOs. As discussed above, building the endogenous elements abstract representation is challenging due to the large number of distinct and varied entities, concepts, functionality, rules and regulations, etc. inside the CNOs. In addition to a variety of tangible elements and resources inside the CNO, there are always networks of organizations in which each node plays a specic role and has heterogeneous relation-

Surrounding environment

CNO

Exogenous Interactions

Endogenous Elements

Fig. 4 Two modeling perspectives

ships with other nodes. Furthermore, there are certain rules of behavior that either constitute the norms, or shall be obeyed by the CNO participants, and needless to say that in every CNO there are a set of activities and functionalities that also need to be abstracted. The four dimensionsE1-Structural, E2Componential, E3-Functional, E4-Behavioralintroduced previously are adequate to model the CNO from the inside perspective. Exogenous interactions This perspective aims at reaching an abstract representation of the CNO as seen from the outside, i.e. which characteristic

123

J Intell Manuf (2007) 18:529542

539

properties the CNO reveals in its interaction with its logical surrounding environment. The purpose here is not to model the surrounding environment but focus on the interactions between the CNO and this environment. A CNO as a whole might interact with, inuence, and be inuenced by a number of interlocutors, e.g. customers, competitors, external institutions, potential new partners. The interactions between the CNO and these external entities are quite different, the same as the way each of these entity groups looks at the CNO. In order to better characterize these interactions, the following additional modeling dimensionsI1-Market, I2-Support, I3-Society, I4-Constituencyare proposed for the external or Exogenous Interactions perspective: Market dimension. This dimension covers both the issues related to the interactions with customers (or potential beneciaries) and competitors. The customers facet involves elements such as the transactions and established commitments (contracts), marketing and branding, etc. On the competitors side issues such as market positioning, market strategy, policies, etc. can be considered. The purpose/mission of the CNO, its value proposition, joint identity, etc. are also part of this dimension. Support dimension. Under this dimension the issues related to support services provided by third party institutions are to be considered. Examples include certication services, insurance services, training, external coaching, etc. Societal dimension. This dimension captures the issues related to the interactions between the CNO and the society in general. Although this perspective can have a very broad scope, the idea is to model the impacts the CNO has or potentially can have on the society (e.g. impact on employment, economic sustainability of a given region, potential for attraction of new investments) as well as the constraints and facilitating elements (e.g. legal issues, public body decisions, education level) the society provides to the CNO development. Constituency dimension. This perspective focuses on the interactions with the universe of potential new members of the CNO, i.e. the interactions with those organizations that are not part of the CNO but that the CNO might be interested in attracting. Therefore, general issues like sustainability of the network, attraction factors, what builds/provides a sense of community, or specic aspects such as rules of adhesion and specic marketing policies for members, are considered here. In addition to these perspectives, a CNO model can be dened at multiple levels of abstraction (model intent perspective). Currently three levels are being considered in our framework:

General concepts levelthat includes the most general concepts and related relationships, common to all CNOs independently of the application domain. Specic modeling levelan intermediate level that includes more detailed models focused on different classes of CNOs. Implementation modeling levelthat represents models of concrete CNOs.

CNO-life-cycle perspective In a typical (long-term) organization, usually its operation stage constitutes its entire livelihood. In other words most successful organizations spend only a negligible fraction of their life time in their setting up and dissolution. Therefore, earlier research on reference modeling for enterprises did not need to elaborate much on the life cycle perspective. But unlike typical organizations, for a wide variety of classes of CNOs their creation stage (as well as their dissolution or metamorphosis) is complex and takes up considerable effort. This is certainly not a negligible fraction of time, and due to the involved complexity, it requires receiving proper attention during the build up of the reference model. Our earlier study of the life cycle stages for CNOs has revealed 4 main stages for the CNO life cycleCreation, Operation, Evolution, and Metamorphosis/Dissolution (Fig. 5). The creation stage can be divided into two phases, namely (i) initiation and recruiting, dealing with the strategic planning and initial incubation of the CNO; (ii) foundation, dealing with the constitution and start up. A grasping-opportunity type of CNO such as a VO will typically dissolve after accomplishing its goal. In the case of a long-term alliance such as a VBE, considering its valuable bag of assets gradually collected during its operation, its dissolution is a very unusual situation. Instead, it is much more probable that this CNO goes through another stage, that we call the metamorphosis stage, where it can evolve by changing its form and purpose. The ongoing plan for this work is to dene A Reference model for COllaborative Networks (ARCON). The elaboration of a comprehensive modeling framework, integrating all the above perspectives, is the rst step of this initiative.

Metamorphosis Creation
Initiation Foundation

Operation

OR

Dissolution Evolution

Fig. 5 CNO life cycle stages

123

540 Fig. 6 Crossing CNO life cycle and the endogenous elements perspective

J Intell Manuf (2007) 18:529542

e.g. e.g. e.g. -Hardware/ software res. -Human res. -Information/ knowledge res. -Ontology res. e.g. -Processes -Auxiliary processes -Procedures -Methodologies

CNO-Life-Cycle Stages

L4. Metamorphosis / Dissolution

-Participants -Relationships -Roles -Geo. location time/culture

-Prescriptive behavior -Obligatory behavior -Constraints & conditions -Contracts & agreements -Incentives

L3. Evolution

L2. Operation

** *

L1. Creation

Endogenous Elements
E1. Structural E2. Componential E3. Functional E4. Behavioral

Inside view

Fig. 7 Crossing CNO life cycle and Exogenous Interactions perspective

e.g.

CNO-Life-Cycle Stages

L4. Metamorphosis / Dissolution

-Customer -Mission -Transactions -Marketing / branding -Market strategy

e.g. -Certification -Insurance -Coaching -Training

e.g. e.g. -Impacts -Legal issues -Public interactions -Attracting factors -Rules of adhesion -Sustainability factors

L3. Evolution

L2. Operation

** *

L1. Creation

Exogenous Interactions
I1.Market I2. Support I3. Societal I4. Constituency

Outside view

Figures 6 and 7 combine the life-cycle perspective with the endogenous elements and exogenous interactions perspectives, respectively. Figure 8 combines the addressed perspectives into a single diagram that summarizes the various perspectives considered in the proposed modeling framework. Brief comparison with other approaches In the investigation and denition of the proposed modeling framework for ARCON, several relevant previous approaches introduced by other initiatives were considered, and certainly contributed to this work (Noran, 2003; Tolle & Bernus, 2003). Our conclusion of this study showed that although several related previous works have provided valuable contributions to the understanding of several aspects of this area, they are somewhat limited when a more holistic modeling is pursued. As an illustration, Table 7 summarizes our analysis results for some relevant initiatives in comparison with the

needs identied for the ARCON reference modeling framework. Clearly, the paper does not argue about the relevance and appropriateness of previous well known models, e.g. Zachman (Zachman, 1987), CIMOSA (Vernadat & Kosanke, 1992), SCOR (Huan, 2004), etc. for the purpose of Enterprise modeling. Rather it argues and represents (i.e. through identifying more elements) that more needs to be done for the purpose of proper Collaborative networked organizations (CNO) modeling. As such, instead of starting from scratch (re-invention of the wheel) to identify the main perspectives and required elements for CNO reference modeling, we have tried to as much as possible reuse some already dened concepts that are more familiar to the users of reference models in this area. Conclusions The establishment of a comprehensive modeling framework for CNOs is a very important basis for the elaboration of a

123

J Intell Manuf (2007) 18:529542 Fig. 8 A modeling framework for CNOs

541

ol ss Di

n io ut

o Ev

ion lut

CNO-Life-Cycle Stages

n tio era Op

l de Mo
on l ra ati ne ent Ge res p Re

nt nte
g ific ec elin Sp od M

n tio nta ing me del ple Mo Im

ea Cr

n tio

Endogenous Elements

Exogenous Interactions

Table 7 Brief summary analysis of other modeling frameworks Approach Zachman Framework (1987) Modeling Framework + Good set of dimensions different emphasis (e.g. location) not clear IN/ABOUT (mixed) - Confusing rows (levels, life cycle, actors) SCOR (Huan, 2004; Stewart, 1997) + Simple composition (client-supplier) and multi-abstraction level -Limited in terms of generality VERAMa (IFIP-IFAC TFAEI, 2003) + Good set of dimensions and abstraction levels Limited life cycle - Confusing/complex generality EGA (2005) + Inclusion of glossary - Grid enterprise infrastructure Manufacturing CNOs PVC and other CNOs) (no Only supply chain (no other CNOs) Modeling Target Single enterprise (not CNOs) Modeling Scope + Most modeling dimensions - Little focus on behavior within the endogenous elements - Little focus on Exogenous Interactions (some aspects in Purpose) -Can cover only process sub-dimension & performance indicators in endogenous elements. - Not addressing Exogenous Interactions + Most modeling dimensions -Little focus on behavior within the endogenous elements -Little focus on Exogenous Interactions - Only part of the functional dimension within the endogenous elements - Not addressing the Exogenous Interactions Governmental organizations + Some aspects Exogenous Interactions - Only process sub-dimension within the endogenous elements

Multiple levels of abstraction -Very limited in terms of generality FEA (2005) + Inclusion of glossary + Customer orientation - Limited in terms of generality
a VERAM

includes elements from PERA (Williams, 1994), CIMOSA (Vernadat & Kosanke, 1992), GERAM (IFIP-IFAC TFAEI GERAM, 2003)

123

542

J Intell Manuf (2007) 18:529542 Camarinha-Matos, L. M., & Afsarmanesh, H. (2006a). A modeling framework for collaborative networked organizations, Network-centric collaboration and supporting frameworks, IFIP (Vol. 224). Boston: Springer. Camarinha-Matos, L. M., & Afsarmanesh, H. (2006b). Towards a reference model for collaborative networked organizations. Information technology for balanced manufacturing systems, IFIP (Vol. 220). Boston: Springer. EGA, Enterprise Grid Alliance Reference Model, 13 Apr 2005. http://www.gridalliance.org/en/workgroups/ReferenceModel.asp Eschenbaecher, J., & Ellmann, S. (2003). Foundation for networking: A theoretical view on the virtual organization, in processes and foundations for virtual organizations. Boston: Kluwer. FEA. FEA Consolidated Reference Model, May 2005, http://www.whitehouse.gov/omb/egov/documents/CRM.PDF. George, V., & Vaughn, R. (2003). Application of lightweight formal methods in requirement engineering. CrossTALK - The Journal of Defense Software Engineering, 16, 30 Huan, S. H. (2004). A review and analysis of supply chain operations reference (SCOR) model. Supply Chain Management: An International Journal, 9(1),2329. IFIP-IFAC TFAEI, GERAM. (2003). The generalized enterprise reference architecture and methodology, IFIP-IFAC task force on architectures for enterprise integration. In P. Bernus, L. Nemes, & G. Schmidt (Eds.), Handbook on enterprise architecture. Heidelberg: Springer. Liu, S. (2005). Formal engineering method for software development an introduction to SOFL. Retrieved February, 2006, from the World Wide Web: http://cis.k.hosei.ac.jp/sliu/FMlecture1.pdf. Lh, H., Zhang, C., & Katzy, B. (2005). Modeling for virtual organizations, in virtual organizations systems and practices. Boston: Springer. Noran, O. (2003). A mapping of individual architecture frameworks (GRAI, PERA, C4ISR, CIMOSA, ZACHMAN, ARIS) onto GERAM. In P. Bernus, L. Nemes, & G. Schmidt (Eds.), Handbook on enterprise architecture. Springer. Stewart, G. (1997). Supply-chain operations reference model (SCOR): The rst cross-industry framework for integrated supply-chain management. Logistics Information Management, 10(2), 6267. Tolle, M., & Bernus, P. (2003). Reference models supporting enterprise networks and virtual enterprises. International Journal of Networking and Virtual Organisations, 2(1), 215. Vernadat, F., & Kosanke, K. (1992). CIM-OSA: A reference architecture for CIM. In Proceedings of the IFIP TC5/WG5.3 eight international PROLAMAT conference on human aspects in computer integrated manufacturing, FIP Transactions; Vol. B-3, (North-Holland). Williams, T. J. (1994). The purdue enterprise reference architecture. Computers in Industry, 24(23), 141158. Zachman, J. A. (1987). A framework for information systems architecture. IBM Systems Journal, 26(3), 276292.

reference model, a base element in the consolidation of existing knowledge in this area, and a basis for its consistent further progress. As a contribution in this direction, a modeling framework for CNOs considering multiple perspectives is proposed. In addition to a framework it is also important to consider the theories, modeling approaches, and tools that better apply to each modeling perspective and dimension. Therefore a comprehensive list of contributions from other disciplines is introduced and mapped into the necessary modeling dimensions for CNOs. Nevertheless it is clear that the establishment of a reference model, able to capture the variety of CNOs and their complexity, is a long term endeavor that needs to start with a careful analysis of the current baseline and denition of related reference modeling frameworks. Current work is focused on the identication of the general modeling elements according to the proposed framework. This analysis is based on different classes of CNOs, namely VO breeding environments, virtual organizations, professional virtual communities, and virtual learning communities.
Acknowledgements . This work was funded in part by the European Commission through the ECOLEAD project. The authors thank the contributions of their partners in this project.

References
Afsarmanesh, H., & Camarinha-Matos, L. M. (2005). A framework for management of virtual organizations breeding environments. Collaborative networks and their breeding environments. Boston: Springer. Alexander, P. (1996). Best of both worlds: Combining formal and semiformal methods in software engineering. Potentials, IEEE, 14(5), 2932. Bowen, J. P., & Hinchey, M. G. (2006). Ten commandments of formal methods... ten years later. Computer, 39(1), 4048. Camarinha-Matos, L. M., & Abreu, A. (2003). Towards a foundation for virtual organizations. In Proceedings of business excellence 2003 int. conf. performance measures, bench-marking, and best practices in new economy. Guimares, Portugal, 1013 Jun 2003. Camarinha-Matos, L. M., & Afsarmanesh, H. (2004). Formal modeling methods for collaborative networks. Collaborative networked organizations a research agenda for emerging business models, cap. 6.3, Springer, Boston.

123