Contents lists available at ScienceDirect

Information Systems

Information Systems 54 (2015) 235–262

http://d0306-43

n CorrE-m

m.e.iacom.j.vansl.ferreir

journal homepage: www.elsevier.com/locate/infosys

Modeling resources and capabilities in enterprise architecture:A well-founded ontology-based proposal for ArchiMate

Carlos L.B. Azevedo a,b,n, Maria-Eugenia Iacob c, João Paulo A. Almeida a,Marten van Sinderen b, Luís Ferreira Pires b, Giancarlo Guizzardi a

a Ontology and Conceptual Modeling Research Group (NEMO), Federal University of Espírito Santo (UFES), Vitória, ES, Brazilb Services, Cybersecurity and Safety Research Group, Centre for Telematics and Information Technology, University of Twente, Enschede,The Netherlandsc Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, The Netherlands

a r t i c l e i n f o

Available online 23 May 2015

Keywords:CapabilityResourceEnterprise architecture modelingOntology-based semanticsArchiMate

x.doi.org/10.1016/j.is.2015.04.00879/& 2015 Elsevier Ltd. All rights reserved.

esponding author.ail addresses: clbazevedo@inf.ufes.br (C.L.B. Ab@utwente.nl (M.-E. Iacob), jpalmeida@ieeeinderen@utwente.nl (M. van Sinderen),apires@utwente.nl (L.F. Pires), gguizzardi@ac

a b s t r a c t

The importance of capabilities and resources for portfolio management and businessstrategy has been recognized in the management literature. Despite that, little attentionhas been given to integrate the notions of capabilities and resources in enterprisearchitecture descriptions. One notable exception is a recent proposal to extend theArchiMate framework and language to include capability and resources and thus improveArchiMate's coverage of portfolio management. This paper presents an ontologicalanalysis of the concepts introduced in that proposal, focusing in particular on theresource, capability and competence concepts. We provide an account for these conceptsin terms of the Unified Foundational Ontology (UFO). The analysis allows us to identifysemantic issues in the proposal and suggests well-founded recommendations forimprovements. We revise the proposed metamodel in order to address the identifiedproblems, thereby improving the semantic clarity and usefulness of the proposedlanguage extension. Two real-world cases are modeled with the resulting metamodel toshow the applicability of the constructs and relations in an industrial setting.

& 2015 Elsevier Ltd. All rights reserved.

1. Introduction

A fundamental question in the field of strategic manage-ment is how organizations gain and sustain competitiveadvantages [8]. Constant shifts in markets, competition,technology and regulation drive organizations topromote changes, and to continuously adapt and improvetheir organizational structures. Failing to address theseshifts affects organizational performance negatively [3].Research has shown that the average period during which

zevedo),.org (J.P.A. Almeida),

m.org (G. Guizzardi).

organizations are able to sustain their competitive advan-tage has decreased over time [74]. This suggests thatorganizations have to build successive temporary advan-tages [17]. In order to be able to accomplish this, anorganization needs to employ its resources and capabilitiesin the most effective manner. This observation is not onlyvalid at the level of one organization, but also withinbusiness networks, in which organizations establish part-nerships with the goal of pooling their capabilities andresources. The suggestion that organizational resources andcapabilities are key success factors for competitive advan-tage has been proposed already in the 90s [7] and is still apredominant topic [37]. A major concern of strategicmanagement is the usage of organizational capabilitiesand resources, in order to improve performance, qualityand to reduce costs [6,58].

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262236

In the work presented in [48], many organizations havebeen analyzed in order to answer the question “to whatextent do access and changes to resource bases influencethe development of dynamic capabilities in new firms?”The work provides statistical evidence for the relationshipbetween organizational resources and the subsequentcapabilities of the organization. Different resources leadto different capabilities, and the changes of resources overtime have a great impact on organizations' capabilities[48].

Strategic management deals with survival and competi-tiveness in the long-term, despite unknown facts that willtake place in the future. This has led to the formulation ofmultiple theories, with a focus on resources [7] and capabil-ities ([17,69]) as a source of competitive advantage.Resources-centric theories regard an organization as a bun-dle of resources [25]. They suggest that the resources'properties (e.g., rare, valuable, non-substitutable or inimita-ble) confer organizations competitive advantage [56]. Theidea is that enterprises with appropriate resources should beable to leverage the needed capabilities and to sustaincompetitive advantages regardless of scenario. Since theintroduction of the resource-based theories, several majorlimitations have been identified. Amongst these, the mostrelevant are the fact that valuable, rare, inimitable, and non-substitutable (VRIN) resources [7] are neither necessary norsufficient for sustaining competitive advantage in a dynamicenvironment [42]. Resources by themselves are not usefulunless they are correctly employed. The way resources areused defines the outcome: [55] stated that “exactly the sameresources when used for different purposes or in differentways and in combination with different types or amounts ofother resources provide a different service or set of services”.As a response to this criticism, the capability-based theorieshave taken shape. According to these theories, the enterpriseneeds to know the capabilities it wants to leverage in orderto use and plan to acquire resources in an intended manner.Whereas resource-based theories focus on accumulatingresources, the capability-based theories focus on “adapting,integrating, and re-configuring internal and external organi-zational skills, resources, and functional competences towarda changing environment” [69].

The need for capability-based planning in the contextof organizations has become more apparent in the recentyears. Recent developments have identified capabilities asthe way to link business and IT [14,68], to link businessoutcomes to IT [49], and as a solution for improving thebusiness and IT alignment ([45,76]).

Given this increased interest in capabilities, both intheory and in practice, it comes as no surprise that thisconcept has recently surfaced in enterprise architecture(EA) following the need to align strategic decisions withtheir actual implementation at the level of processes, ITsystems and infrastructure. EA frameworks such as TOGAF[70] have already introduced basic notions of capability-based planning and its role in designing, planning andimplementing organizational change.

One of the main issues addressed by capability-basedplanning, in the context of EA, is to provide enterprisearchitects with a common ground to initiate discussionswith business leaders in terms of business outcomes

(increased output, better quality, lower costs, revenuegrowth or improved market share) instead of projects,processes and applications [52]. The problem is thatprocesses are too detailed, applications are too technicaland projects focus on short term results, usually havinglittle strategic value. Despite the relevance of capabilitiesand resources to the success of enterprises, little attentionhas been given to integrate these notions of capability andresource in enterprise architecture descriptions. A notableexception is the work discussed in [41], which extends theArchiMate language [40] with constructs to representcapabilities and resources, integrating these new con-structs with those used to represent other aspects of anenterprise architecture (such as active structure and beha-vior). The objective is to empower enterprise architects touse these important notions in coherent enterprise archi-tecture descriptions. We argue that modeling resourcesand capabilities for decision making purposes at strategiclevel must simplify models and hide the complexity ofarchitecture models which is of no relevance at thatabstraction level, where decision makers are mostly inter-ested in means (i.e., resources and capabilities) and goals(i.e., motivation). This is due to the fact that both capabil-ities and resources are defined in the proposed extensionas abstractions of complex behavior and architecturalstructure, respectively. This is also why important benefitsto be reaped from modeling resources and capabilities areat the strategic level, while carrying out activities such asstrategic alignment, capability-based planning, capabilityportfolio management, etc. In contrast, resources andcapabilities can be linked to the architecture fragmentsthey are abstracted from, thus enabling the end-to-endtraceability from strategic decisions to implementationand architecture change.

A main challenge of incorporating the notions of cap-ability and resource in enterprise architecture (EA) lies inidentifying a precise conceptualization for these notions.Without such a precise conceptualization, rigorous defini-tion of the semantics of the proposed modeling elementsis problematic, and modeling and communication pro-blems arise. For example, when various modelers share amodel without a clear semantics, False Agreement [27]most likely ensues. In that case, different modelers cometo different interpretations of the same model and are notaware of the conflict. This would result in enterprisearchitecture models that cannot serve their purpose astools for communication between stakeholders, decreasingthe value of enterprise architecture models in the pursuitof informed decision-making.

Our objective in this paper is to address this challenge,resulting ultimately in a well-founded approach to capturecapabilities and resources in enterprise architecturedescriptions. The novelty of our approach resides in thestrong relation of enterprise architecture modeling tothese concepts as introduced in strategic management,and, most importantly, in the clear semantics we providefor the proposed modeling elements through ontologicalanalysis. We aim at addressing the strategic managementlevel, allowing capability-based planning and better deci-sion making. To accomplish that, we analyze the semanticsof the Business Strategy and Valuation Concepts (BSVC)

Table 1Concrete syntax and definitions for resource, capability and competenceas it is on the BSVC.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 237

extension to ArchiMate [41] building up on our earlierwork presented in [4]. We identify issues with the originalproposal, by interpreting it in terms of a reference founda-tional ontology. We then introduce a new ontologicallywell-founded version of the BSVC metamodel thataddresses the identified issues.1

The remainder of this paper is structured as follows:Section 2 describes the current modeling of resources andcapabilities in EA using the ArchiMate Business Strategy andValuation Concepts extension as our starting point andintroduces a running example. Section 3 introduces ontolo-gical analysis, in order to clarify the methodology we employ.Section 4 presents the ontological foundation concepts usedin our analysis. Sections 5–7 provide interpretations for theconcepts of resource, capability and competence, respec-tively. Each concept is interpreted according to the metamo-del, the textual definitions, and the examples of usageprovided in [41], and in the related literature. Section 8proposes a number of improvement recommendations,including a new metamodel proposal for the language.Section 9 illustrates the proposed extension and recommen-dations by means of two use cases: the Toyota case, whichfocuses on the conceptual modeling aspects, and the ITconsolidation case, which is an example of capability-basedplanning, showing how the strategic decision of facilitatingand improving the switching of customers (as result of an EUregulation) has led to an IT consolidation problem for whichthe implementation solution is based on the selection of anoptimal portfolio of systems. Section 10 discuses relatedwork. Finally, Section 11 presents our conclusions and out-lines future work.

2. Modeling capabilities, resources and competences

Organizational resources and capabilities are key fac-tors for organizational success. Our goal is to enablemodelers to capture capabilities, resources and compe-tences in an enterprise architecture modeling language.We use ArchiMate as a starting point.

The modeling of resources and capabilities for Archi-Mate was proposed in [41]. The Business Strategy andValuation Concepts extension was based on the analysisand review of relevant business strategy and portfoliomanagement literature. For this reason, many conceptspresent in other approaches, such as [24,75], have beenincorporated in the proposed extension.

Similarly to other earlier ArchiMate extension propo-sals (e.g., the motivation extension presented in [5]) theinitial development of the BSVC has been conductedwithout a rigorous definition of the semantics of theproposed modeling elements. The absence of such defini-tions could lead to several modeling and communicationproblems. Further, this would result in enterprise archi-tecture models that cannot serve their purpose as tools forcommunication between stakeholders, and woulddecrease the enterprise architecture value in the pursuitof informed decision-making.

1 This paper extends [4] in which the issues with the originalproposal were identified.

The BSVC introduces to ArchiMate the concepts ofresource, capabilities, competence and risk. We focus onthe resource, capability and competence concepts in thispaper. The definitions and concrete syntax for the model-ing elements introduced by the extension are shown inTable 1. In this section we have preserved the definitionsas they were provided originally in [41]. These are theobjects of analysis in Sections 5–7.2

Fig. 1 shows the metamodel fragment, as proposed in[41], for the integration of the BSVC with the ArchiMatecore metamodel and its extensions.

The resource concept is prominently present in mostvaluation techniques, in business modeling approaches,and in constraint optimization models in which they aremathematically defined and constrained. This supports theimportance given to the resource concept in the manage-ment literature [41].

A resource represents an asset owned or controlled byan individual or organization. Resources are related to themotivation extension, in particular to requirements andgoals, through the realization relation. The argument forthis relationship is that goal achievement assumes avail-ability and (constrained) consumption of certain resources.This view is based on the mathematical formulation ofconstrained optimization models, in which a goal functionis minimized/maximized subject to a system of constraints(expressed as inequalities) imposed on the resources to beconsumed for the achievement of the goal. Thus, aresource may satisfy a requirement, which in turn, mayrealize a goal. Furthermore, a resource is realized bystructure elements, and is regarded in [41] as an abstrac-tion of these elements.

Capability is defined as the ability (of a static structureelement, e.g., actor, application component, etc.) to employresources to achieve some goal. Similarly to resource,capability is regarded in [41] as an abstraction of somebehavior. The assignment relationship between resourceand capability expresses the ability to employ (i.e., config-ure, integrate, etc.) resources.

The competence concept is introduced in [41] as aspecialization of resource, based on the definition of

2 The competence element has no individual concrete syntax and isrepresented by the resource concrete syntax in [41].

Fig. 1. BSVC Metamodel fragment [41].

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262238

competence proposed by [57], which equates competenceto personnel-based resources.

Fig. 2 shows an ArchiMate model that was defined beforethe ontological analysis to reflect the range of elements andrelations used to represent capabilities and resources in theproposed ArchiMate extension. The example focuses on theArchiSurance company and its “Insuring Capability”. Theexample shows other capabilities that are associated to the“Insuring Capability” (“Claim Handling Capability”, “SellingCapability”, “Damage Assessment Expertise Capability” etc.).The example also reviews the resources assigned to thecapabilities (“Damage Assessment Resources”, “Money”,“Authorized Garage”, etc.). The example reflects that capabil-ities can be decomposed into other capabilities, and can berealized by some behavior elements (as in the case of“Insuring Capability” and “Claim Handling Capability” beingrealized by the “Assess claim” business process, by the“Financial administration” business function and by the“Claim registration” application service). Similarly, resourcescan be realized by structure elements and resources can beassigned to capabilities (as in the “Car Damage AssessmentResource” and “Claim Handling Capability” case).

3 Adapted from [63].

3. Ontological analysis

Since the last 80s there has been a growing interest inthe use of foundational ontologies for evaluating andreengineering conceptual modeling languages and meth-odologies, as in [71,72]. A foundational ontology defines asystem of domain-independent categories and their ties,which can be used to articulate the conceptualizations ofreality. The use of foundational ontologies aims to ensure

ontological correctness of the language and of the modelsdescribed with the language.

Empirical evidence [11,23,60,67] has corroborated thehypothesis that a suitable conceptual modeling languageshould comprise modeling elements that reflect concep-tual modeling categories and relations defined in a foun-dational ontology. A number of enterprise modelingapproaches have been subject to ontology-based analysisin recent years, e.g., [5,15,19,26,33,43,44,61,63].

Ontological analysis is performed by considering amappingbetweenmodeling constructs and the concepts in an ontology.“On one hand, each modeling element can be interpretedusing the ontological theory as a semantic domain. On theother hand, concepts of the domain of discourse (captured inthe ontological theory) should be represented by modelingelements of the language being considered. According to [73],there should be a one-to-one correspondence between theconcepts in the ontology and modeling elements” [63]. Whenthe correspondence cannot be obtained, the following lan-guage problems can be identified3:

–

Construct excess: exists when a modeling construct doesnot correspond to any ontological concept. Since nomapping is defined for the exceeding construct, its mean-ing becomes uncertain, hence, undermining the clarity ofthe specification. According to [72], users of a modelinglanguage must be able to make a clear link between amodeling construct and its interpretation in terms ofdomain concepts. Otherwise they will be unable to articu-late precisely the meaning of the unclear construct and,

Fig. 2. Example model using the BSVC extension of ArchiMate.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 239

consequently, the specifications they generate using thelanguage. Therefore, a modeling language should notcontain construct excess and every instance of its model-ing constructs must represent an individual in the domain.

–

Construct overload: exists when a single modelingconstruct can represent multiple ontological concepts.Construct overload impacts language clarity negatively.Construct overload is considered as an undesirableproperty of a modeling language since it causes ambi-guity and, hence, undermines clarity. When a constructoverload exists, users have to bring additional knowl-edge not contained in the specification to understandthe phenomena which are being represented.

–

Construct redundancy: exists when multiple modelingconstructs can be used to represent a single ontologicalconcept. Construct redundancy is a violation of parsi-mony. In [72], the authors claim that construct redun-dancy “adds unnecessarily to the complexity of themodeling language” and that “unless users have in-depth knowledge of the grammar, they may be con-fused by the redundant construct. They might assumefor example that the construct somehow stands forsome other type of phenomenon.” Therefore, constructredundancy can also be considered to underminerepresentation clarity.

–

Construct deficit: exists when there is no construct inthe modeling language that corresponds to a particularontological concept. Construct deficit entails lack ofexpressivity of the modeling language, i.e., there arerelevant phenomena in the considered domain

(according to a domain conceptualization) that cannotbe represented by the language. Alternatively, users ofthe language can choose to overload an existing con-struct, thus, undermining clarity.

Fig. 3 illustrates the relation between modeling con-

structs (in a language's syntax) and ontological concepts.

Modeling languages aiming at wide adoption, such asArchiMate, should avoid ontological deficiencies. Reckeret al. [60] reported results from a study with 528 modelersdemonstrating that “users of conceptual modeling gram-mars perceive ontological deficiencies to exist and thatthese deficiency perceptions are negatively associatedwith usefulness and ease of use of these grammars”.

In order to uncover any of the aforementioned pro-blems, we discuss the interpretation of the resource,capability and competence modeling elements in termsof a foundational ontology. The issues uncovered inSections 5–7 are addressed with recommendations forimprovements in Section 8.

4. Ontological foundations

In our ontological analysis we make use of the UnifiedFoundational Ontology (UFO) as semantic foundation.The UFO ontology has been developed based on theoriesfrom Formal Ontology, Philosophical Logics, Philosophy ofLanguage, Linguistics and Cognitive Psychology [28]. Likeother foundational ontologies, such as DOLCE [20] and GFO(General Formalized Ontology) [38], it has the firm

Fig. 3. illustrates the relation between modeling constructs (in a language's syntax) and ontological concepts.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262240

ontological grounding of the so-called Aristotelian Square(or Four-Category Ontology) [46]. This allows for the con-struction of an ontology that is founded in a parsimoniousset of essential ontological categories, while still being ableto account both for natural science as well as linguistic andcognitive phenomena [28]. However, differently fromDOLCE and GFO, UFO was developed with the primary goalof providing foundations for conceptual modeling. As aconsequence, many aspects that are essential for conceptualmodeling have not received a sufficient attention in DOLCEand GFO. One example is the notion of material relationsand relational properties ([30]), which does not receive anytreatment in DOLCE. Another example is the finer-graineddistinctions that are needed for dealing with differentcategories of universals (e.g., kind, roles, roleMixins) inconceptual modeling, which are not addressed in GFO. Infact, a category such as roleMixin, which is so recurrent inconceptual modeling to the point of becoming a semanticdesign pattern ([32]), is not present in any of these alter-native foundational ontologies but UFO. A third example isthe availability of a rich system of part-whole relations.Part-whole relations have received a richer treatment inUFO than in other foundational ontologies. Furthermore,UFO presents a treatment of dispositions (and their sys-tematic connection to events and situations), which allowsus to properly address many of the issues that are germaneto the purposes of this paper.

The UFO ontology has been employed in many seman-tic analyses, including ones in the topics of ARIS EPCs [64],in/TROPOS ([34]), goals and business processes models[13] and role-related concepts in EA [1]. In particular, it hasbeen previously used to analyze and interpret the seman-tics of the ArchiMate motivation concepts, having led torecommendations of that proposal that have been incor-porated in the latest ArchiMate specification [5].

The UFO ontology has been structured in three main parts:UFO-A is the core of the ontology and is concerned withendurants (e.g., entities such as objects, qualities, relators, anddispositions); UFO-B is an ontology of events and, as such,makes a distinction between enduring and perduring indivi-duals and elaborates on the possible connections betweenthese two fundamental types of entities; UFO-C is built on topof UFO-A and UFO-B and focuses on social aspects of reality by

dealing with notions such as plans, goals, agents, commit-ments, and normative descriptions.

For a more detailed discussion of the development andapplications of UFO we refer to [2,28,29,31]. This section isbased on the UFO description that appeared in [5].

4.1. Basic elements

A fundamental distinction in this ontology is betweenthe categories of individuals and universals. Universals arepredicative terms that can be applied to a multitude ofindividuals, capturing their general aspects. Individuals areentities that exist in reality possessing a unique identityand that can instantiate one or more universals.

Further, UFO makes a distinction between the conceptsof endurants and events. Endurants are individuals thatpersist in time while keeping their identity, in the sensethat if we say that in circumstance c1 an endurant e has aproperty p1 and in circumstance c2 a property p2 (possiblyincompatible with p1), e is the same endurant in each ofthese situations. Examples can include a particular person(say Peter) weighting 70 kg in one circumstance and 78 kgin a different circumstance, while being the same indivi-dual (Peter) in these two circumstances. Other examplesinclude organizations (the University of Twente, the Fed-eral University of Espírito Santo, etc.) and everyday objects(a ball, an apple, etc.). Events, in contrast, are individualscomposed of temporal parts, they happen in time, in thesense that they extend in time and accumulate temporalparts. Examples include a particular execution of a busi-ness process, a meeting or a soccer game. Whenever anevent occurs, it is not the case that all of its temporal partsnecessarily occur. For instance, if we consider a businessprocess “Buy a Product” at different time instants, at eachtime instant only some of its temporal parts are occurring.

A substantial is an endurant that does not dependexistentially on any other individual, what is usuallyreferred by the common sense term “object”. In contrastwith substantials, moments (also known as “abstract parti-culars” and “tropes” [36,46]) are existentially dependententities, i.e., for a moment x to exist, another individualmust exist, named its bearer. Examples of moments includean apple's color, John and Mary's marriage, an electric

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 241

charge on a conductor, etc. Moments in UFO include bothqualities (e.g., color, weight, temperature) and dispositions(e.g., the fragility of a glass, the disposition of a magnet toattract metallic material) [31]. In the philosophical jargon,the category of dispositions typically subsumes propertiessuch as powers, tendencies, potentials, capacities, capabil-ities, affordances, liabilities and propensities. In general,these properties have in common that they endow theirbearers with the potential of exhibiting some behavior orbringing about a certain effect under certain conditions.Dispositions are only manifested in particular situations,but they can also fail to be manifested. When manifested,they are manifested through the occurrence of events.Take, for example, the disposition of a magnet m to attractmetallic material. The object m has this disposition even ifit is never manifested, for example, because m was neverclose to any magnetic material. Nonetheless, m can cer-tainly be said to possess that intrinsic property [36,51,53].

Existential dependence can be used to differentiateintrinsic and relational moments. Intrinsic moments aredependent on a single individual, while relational moments(also called relators) depend on a plurality of individuals.Examples of the first include an apple's weight and color,while examples of the latter include John and Mary'smarriage, John's enrollment at the University of Twente.A relator is the truthmaker of a material relation.

A universal is rigid if it necessarily applies to itsinstances, i.e., if it applies to its instances in every possibleworld (e.g., Apple, Person). A kind is the rigid substantialuniversal that supplies a principle of identity for thesubstantial individuals that instantiate them. Every sub-stantial individual must be an instance of exactly one kind.In contrast to rigid universals, a universal is anti-rigid if itdoes not apply necessarily to all its instances. Roles areanti-rigid and relationally-dependent universals (e.g., Stu-dent, Husband). This means that roles are played by asubstantial in a relational context, i.e., it requires the roleplayer to be connected by a relator to other individuals inthat context or that the role player participates in an eventplaying what is termed a processual role.

Whenever entities of different kinds have similar prop-erties they may be classified by substantial universalstermed Mixin universals. Rigid Mixin Universals subsumingdifferent kinds are termed categories (e.g., Physical Object,Living Entity). Some Mixin Universals are anti-rigid andrepresent abstractions of common properties of roles.These are termed role mixins. An example of role mixin is“customer”, which can be played by “persons” (i.e., entitiesof the kind Person) and “organizations” (i.e., entities of thekind Organization). In any case, customer is an anti-rigidand relationally dependent type for all its players.

4.2. Intentional and social elements

An agent is a specialization of substantial individual,representing entities capable of bearing intentional moments.These include mental states, such as individual beliefs, desiresand intentions. Intentionality should not be understood asthe notion of “intending something”, but as the capacity torefer to possible situations of reality [66]. Every intentionalmoment has an associated proposition that is called the

propositional content of the moment. In general, the proposi-tional content of an intentional moment can be satisfied (inthe logical sense) by situations in reality. Every intentionalmoment has a type (belief, desire or intention). The proposi-tional content of a belief is what an agent holds as true.Examples include one's belief that the Eiffel Tower is in Parisand that the Earth orbits around the Sun. A desire expressesthe will of an agent towards a possible situation (e.g., a desirethat Brazil wins the next World Cup), while an intentionexpresses desired states of affairs for which the agentcommits to pursuing (internal commitment) (e.g., John'sintention of going to Paris to see the Eiffel Tower).

Actions are intentional events, i.e., events with thespecific purpose of satisfying (the propositional contentof) some intention of an agent. The propositional content ofan intention is termed a goal. Only agents are said toperform actions [29], as opposed to non-agentive objectsthat participate (non-intentionally) in events.

Agents can be further specialized into physical agents (e.g., a person) and social agents (e.g., an organization). Socialagents are further specialized into institutional agents andcollective social agents. Institutional agents are composed ofother agents, each one contributing to the functionality (orbehavior) of the institution, also termed functional complex[28]. In addition to institutional agents, UFO also acknowl-edges the existence of collective social agents, which aredistinguished from institutional agents in that all its mem-bers play the same role in the collective.

Similarly to agents, non-agentive objects can be specia-lized into physical objects and social objects. A category ofsocial objects of particular interest to us here is that ofnormative descriptions. Normative descriptions are socialobjects that create social entities recognized in that con-text. Examples of normative descriptions include a com-pany's regulations and public laws. Examples of socialentities that can be defined by normative descriptionsinclude social roles (e.g., president, manager, sales repre-sentative), social role mixins (whose instances are playedby entities of different kinds, e.g., customer, which can beplayed by persons and organizations), social agent univer-sals (e.g., that of political party, education institution),social agents (e.g., the Brazilian Labour Party, the Universityof Twente), social object universals (e.g., currency) andother social objects (e.g., the US dollar) or even othernormative descriptions (e.g., a piece of legislation). Norma-tive descriptions are recognized by at least a social agent.Fig. 4 shows a fragment of the specializations of indivi-duals in UFO.

5. Ontological analysis of resource

In this section we discuss the ontological analysis andinterpretation of the resource modeling element intro-duced in [41]. We discuss possible interpretations in termsof UFO and consider the consequences of the variousinterpretations on the usage of the language. We make athorough analysis, and the section is structured accordingto six problems that have been revealed in this analysis,which for convenience are named R1 to R6. The analysis ofthe revealed problems leads to the recommendationspresented in Section 8.

Fig. 4. Fragment of specializations of individuals in UFO.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262242

5.1. Problem R1

The Oxford Dictionary defines resource as “a stock orsupply of money, materials, staff, and other assets that canbe drawn on by a person or organization in order tofunction effectively”. In [41], a similar intuition is putforward when motivating the resource element in theBSVC extension: “the achievement of a goal assumes theavailability and (constrained) consumption of certainresources”. Further, resources are also characterized as“assets owned or controlled by an organization”.

Since “assets” are (valuable) things, in a first examination,this characterization seems to suggest that resources repre-sent specific individuals, such as business actors (e.g., in caseof staff as resource) or business objects (agentive objects(agents) or non-agentive objects). Nevertheless, this interpre-tation would show a clear case of construct redundancy [28],as the additional resource modeling element would serveno purpose, being supplanted by the previously existingstructure elements of the language (such as business actor andbusiness object). Wemust conclude this is not the intention ofthe designers of the extension, which indicate further that aresource is “an abstraction of structure elements” and includea “realizes” relation that may be used to connect structureelements to the resources they “realize”. This suggests that itis not the specific structure element that is represented usinga resource, but some more abstract notion, which reveals thedependence on a structure element with certain character-istics without specifying the particular element involved. Inother words, we understand that the resource elementdefines some type of structure element (a universal), and thatthe structure element that realizes the resource instantiatesthis type. An example of this would be a model that includesa business actor “John” that realizes a resource called “CarDamage Assessment Resource”.

If we further consider that resources are used in the scopeof the efforts to achieve goals, we can understand that aresource refers to the role an object (agentive or non-agentive)plays when employed in the scope of these efforts. In theexample above, “John” functions as a “Car Damage

Assessment Resource” in some context of the organization(for example, that of the process of assessing damages).

Further, in order to play a particular role, an object maybe required to instantiate some particular type (what iscalled an “allowed type” in [10]). For example, any “CarDamage Assessment Resource” may need to possess spe-cific damage assessment skills, and thus instantiate somespecific universal that is characterized by these skills.

Thus an intermediate conclusion is that resource representsan externally dependent universal (either a role or role mixin)that may be instantiated by objects of a particular allowedtype. The fact that resource models both the role an objectplays in a particular context of usage as well as its allowedtype has some consequences to the terms used in the label ofa resource. In some cases, such as “Car Damage AssessmentResource”, the context of usage is emphasized, focusing thuson the role or role mixin that is instantiated when the resourceis used; in other cases, such as “Money”, the allowed typerequired for role playing is emphasized in the label.

An insight that comes out of this interpretation is that,as a role (mixin), a resource should have a context of usage,which in UFO is defined in the scope of a material relation(or in the scope of an event). This means some asset is aresource in a defined context, but not in others. Forexample, the “Car Damage Assessments Expert” is aresource in the “Damaged cars inspection process” becauseof the “Car Damage Assessment Expertise”, but it shouldnot be considered a resource for the “Collect premiums”process. This cannot be directly represented with the BSVCmetamodel, because there is no notion of “use” ofresources. We label this as problem “R1”.

5.2. Problems R2 and R3

The resource concept is also defined as “an asset owned orcontrolled by an individual or organization”. Being “ownedor controlled” is understood as being available for theorganization, e.g., by an employment contract betweenemployers and employees, or by having the right/ownershipover a certain object. For the cases inwhich the resource is an

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 243

agentive element (agent) we understand the “controlling” inthe context of the social relator that bounds the particularindividual or organizationwith the first, e.g., the employmentcontract. A controls Bmeans that there is a (possibly a set of)meta-commitment(s) of B towards A. In other words, A hasmeta-claims over B and, hence, the ability to delegate to Band, consequently to increase its social ability [12]. For theinterpretations in which the resource is a non-agentive object,we understand “the control or owning of the asset” as theability to have that element to participate in an event ofinterest, in which the organization A has a certain right withrespect to the object O (for example, a right to use, toconsume, to destroy, to sell, etc.). The current metamodeldoes not allow the modeler to identify who controls theresource, aside from the use of the very general and abstractassociated with relation, which has no specific semantics andcan be applied between any constructs in the language. Thiscould be an issue when there are multiple business actors(different organizations, business departments) that couldcontrol this resource. For example, the model user would notbe able to know which organization controls the resourcemodeled, and as such, could not distinguish if that resource isor is not available at his organization. We call the lack ofexpressiveness of control relations “R2”.

Furthermore, the extension does not distinguish betweenresources potentially played by agents from those potentiallyplayed by non-agentive objects. We call this lack of expres-siveness “R3”. Further, R3 has an impact on R2, since non-agentive resources are to be controlled by someone, and, asimportantly, should not control any other element.

5.3. Problems R4 and R5

Other examples of usage indicate that not only businessactors and objects may realize a resource, but also that businessrolesmay be said to realize a resource. In our example, the “CarDamage Assessment Expert” business role realizes the “CarDamage Assessment Resource”. In this case, we should under-stand that whichever object instantiates the role representedby the business role may also instantiate the role (mixin)represented by the resource. Intuitively in the example, notonly “John” but also any other damage assessment expert is a“Car Damage Assessment Resource”. For these cases, thelanguage does not determine whether one or more individualsinstantiating the role (mixin) represented by the resource arerequired, used or controlled in the particular context. In otherwords, it is not possible to express whether all the instances ofthat type are required, used or controlled, if just one instanceof that type is required, used or controlled or if an arbitrary setof instances of that type are required, used or controlled. Welabel this as problem “R4”.

Further, the current ArchiMate language does notaddress the cases in which the resources are objects ofinterest or raw materials (e.g., “Money” as in the runningexample, or gold, diamond, gas), i.e., passive non-agentiveelements. We believe the language designers have tried tocover this by stretching the resource element, and using itdirectly to represent such objects. However, these wouldbe resources that do not have any structure element torealize them, since there are no structure elements that canrepresent these types of objects in ArchiMate (the passive

structure of ArchiMate focuses primarily on informationobjects). Also, in these cases, the resources are role (mixins)and the language is not able to express if the sameinstance is to be considered a resource in various contexts(e.g., usage of the same amount of money in differentcontexts). Also, it is not possible to express any propertyassociated to the element itself (e.g., quantity of money,gold carats). We label this as problem “R5”.

The resource concept is related to the motivation extensionthrough the realization relation, inwhich a “resource realizes arequirement”. According to [5], a requirement corresponds to anormative description, which states that if a system (in a broadsense) is to exist, then it must satisfy a particular proposition.In this case, we understand that proposition refers to the object(or objects) playing the resource role. Any instance of the role(mixin) represented by the resource must satisfy the require-ment's proposition. To put it simply, a requirement addscharacteristics to a resource's allowed type.

The proposal also states that “the achievement of a goalassumes the availability and (constrained) consumption ofcertain resources”. However, goals are not associateddirectly to resources, and the proposal is silent on the issueof resource consumption. Resources and goals are onlyindirectly related through the “goal is realized by require-ment” and the “requirement is realized by resource” rela-tions. At this point, no interpretation can be given to thetextual definition, and further language documentationwould be required on the topic of resources availabilityand consumption.

5.4. Problem R6

Now we focus on the common ArchiMate relationshipsthat apply to the resource concept (specialization andaggregation). We interpret the specialization relationbetween resources as subsumption between the roles or(role mixins) represented by the resources. The aggregationrelation between resources suggests some sort of whole-part relationship, since aggregation in ArchiMate may berepresented by containment (see in Fig. 2 the relationbetween “Damage Assessment Resources” as a whole, and“Authorized Garage”, “Damage Assessment Team” and“Car Damage Assessment Resource” as parts). Sinceresources may represent both agentive and non-agentiveobjects, it would be possible to combine these withaggregation. We interpret this as a very general sort ofwhole-part relation known as mereological sum. However,there is no distinction between AND or OR resourceaggregations in ArchiMate. Thus, when resources areaggregated, it is unclear whether all the aggregatedresources are required/used, or whether one or any arbi-trary number of them is required. We label this asproblem “R6”.

6. Ontological analysis of capability

In this section we discuss the ontological analysis andinterpretation of the capability modeling element intro-duced in [41]. This section discusses the 5 problemsrevealed by the ontological analysis, named C1 to C5,although the ontological analysis is continued throughout

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262244

the subsections. The revealed problems inspired therecommendations in Section 8.

6.1. Problem C1

The BSVC proposal defined capability as “the ability (ofa static structure element, e.g., actor, application compo-nent, etc.) to employ resources to achieve some goal. […]Also capability assumes the ability to employ (i.e., config-ure, integrate, etc.) resources”. The definition also statesthat “capability (similarly to resource) can be seen as anabstraction of some behavior of the static structureelement”.

We intuitively understand that a capability is attributedto some agent and gives that agent its power to bring aboutsome behavior in order to achieve a desired outcome.From the excerpt “of a static structure element, e.g., actor,application component, etc.”, the capability appears tobelong to the specific individual that is to bring about thedesired outcome. This would lead us to interpret thatcapabilities are dispositions in UFO (dispositions, i.e.,moments that are only manifested in particular situationsand that can also fail to be manifested). However, carefullyexamining the “abstraction of some behavior” fragment,and considering the same pattern that was employed bythe language designers with respect to resources (as“abstractions of structural elements”), we understand thatcapabilities should be interpreted as types of dispositions(disposition universals in UFO).

Often a capability represents a general disposition type.For example, the “Car Damage Assessment Expertise”capability is a general disposition type that is implicitlyspecialized into a more specific type (e.g., the capabilitiesto assess car damage produced by fire, to assess cardamage caused by flood, and to assess car damage aftera crash).

This dispositional account is also applicable for cases inwhich some organization might hire a different organiza-tion to perform processes that realize a capability, and stillstate to have that capability, in this case because it has thedisposition of delegating the capability [12,34]. This isrelated to the idea of what an organization can “sociallyperform”. If A has a meta-commitment from B to execute S,then A (socially) can do S. An object can have dispositionswhich arise from its parts, or from the network of itsdelegation relations [12].

Preferably, the language should allow us to infer whichindividuals bear the dispositions that are related to thatcapability. However, the original metamodel does notinclude relations between the capabilities and the structureelements that are said to have the capabilities, such as abusiness actor or business role. As a consequence, thelanguage does not allow one to identify the individuals ortypes of individuals that bear dispositions of the typepresented by the capability, aside from using the genericand semantically-neutral association of ArchiMate. In otherwords, it is not possible to express in the language whichstructure element has a capability, including the capabilitiesan organization has, unless the capability is realized bysome behavior element. We label this as “problem C1”.

For the case of resources, these are assigned to acapability, in the sense that they are used in order toleverage capabilities, but resources themselves do not havecapabilities in the original extension. This is represented inFig. 2, in which the organization has not assigned the“selling capability” to any resource. Since the languagecannot express which are the capabilities inhering in the“Salesman” resource and the “Sales Manager” structureelement (business role), the organization is not able toknow which resource or structure element has the required“selling capability” to properly assign its performance.

6.2. Problems C2, C3 and C4

The capability concept has three defined relationships,according to the original metamodel. We now focus on“capability realizes requirement”. Again, according to [5], arequirement corresponds to a normative description, whichstates that if a system (in a broad sense) is to exist, then itmust satisfy a particular proposition. In this case, weunderstand that proposition refers to the dispositions thatinstantiate the disposition universal represented by thecapability. The dispositions must be in accordance withthe requirement, in order to satisfy its proposition.

We now focus on the relations “capability realized bybehavior element” and “resource assigned to capability”. Weunderstand that the first needs to be considered also withthe participation of the resource (via the “resource assignedto capability” relationship). We understand that this pat-tern of relations can have two different interpretations. Welabel this as “problem C2”. A first one is that the resourceobject has a disposition that instantiates the dispositionuniversal represented by the capability, and that theparticipation of the resource manifesting its disposition isrequired in order to perform the behavior element (an eventuniversal). For example, the “Car damage AssessmentExpertise” capability on Fig. 2 is to be manifested in the“Damaged cars inspection process” business process inorder for the organization to perform that process. Thesecond possible interpretation is that the capability isacquired (by the resource) with the performance of thebehavior element [51], i.e., the resource acquires a capabilityafter the process is performed. For example, the “Cardamage Assessment Expertise” capability is acquired withthe occurrence of the “Damaged cars inspection process”process. In UFO, this can be interpreted as follows: s is asituation in which the object has the disposition d, e is anevent representing the behavior element, and e is a pre-stateof s. If no resource is represented, the object that isacquiring or manifesting the capability is unknown. Welabel this as “problem C3”. In this case, one can argue thatit is an organization's capability, but it is not possible toclearly define it without relating it to the object, and thespecification does not define this case. Even when relatedto the resource concept, since the resource that is acquiringthe disposition represents a universal, a type element, theactual object that is acquiring the disposition is undeter-mined. The language is not expressive enough to state ifone individual, all the individuals that instantiates theuniversal or an arbitrary combination of individuals instan-tiating the universal are acquiring the disposition. In none

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 245

of these interpretations it is possible to know in advance ifit is one individual, all the individuals that instantiates thatuniversal or an arbitrary combination of them that arerelated to the disposition. In the first interpretation, it is notpossible to know how many objects are to manifest theirdispositions in the event represented by the behaviorelement. In the later, it is not possible to know whichobject is to acquire the disposition. We label this as“problem C4”.

6.3. Problem C5

Now we focus on the common ArchiMate relations thatapply to the capability concept. The specialization relationbetween capabilities should mean that a disposition uni-versal (type) subsumes other disposition universal, and theaggregation relation between capabilities is interpreted as(complex) dispositions, which are dispositions based onother dispositions [51]. However, there is no distinctionbetween AND or OR capability aggregations in ArchiMate,i.e., it is not clear whether the “aggregated” capabilities areall required or whether there are optional capabilities. Thelanguage also lacks expressiveness to state if all thecapabilities associated to a behavior element are acquiredor manifested, if just one of them is acquired or mani-fested, or if an arbitrary number of them are acquired ormanifested. We label this as “problem C5”.

7. Ontological analysis of competence

In this section we discuss the ontological analysis andinterpretation of the competence modeling element intro-duced in [41]. This section discusses the 3 problemsrevealed by the ontological analysis, named C1.1, C2.1and C6, although the ontological analysis is continuedthroughout the subsections. The revealed problemsinspired the recommendations in Section 8.

7.1. Problems C1.1 and C2.1

The competence concept was introduced in [41] as a“specialization of resource (intangible or personnel-based)”.The proposal states that “a core competence is a particularstrength of an organization. Core competences are thecollective learning in organizations, and involve how tocoordinate diverse production skills and integrate multiplestreams of technologies. Examples of core competencesinclude technical/subject matter know-how, a reliable pro-cess and/or close relationships with customers and suppli-ers”. This was “based on the fact that the definition ofcompetence […] is almost identical with that of personnel-based resources”. Personnel-based resources have been exem-plified as “technical know-how, other knowledge assetsincluding organizational culture, employee training, […]”.

Based on the aforementioned statements regardingcompetence, we understand that a competence is some-thing that an element, when provoked, is able to do, or toperform. It addresses the element's capacity of performingan activity. At a first examination, this characterizationseems to suggest that competence is to be applied tospecific individuals, and that it would be a disposition that

inheres in the individual. However, the original metamodelshows that competence is a resource, i.e., a universal. Basedon the resource interpretation as a universal, we concludethat the competence also represents a universal, whichwould be a disposition type, whose instances inhere inthe objects that play the role represented by a resource.

Since a competence is a specialization of resource in theBSVC, it inherits the resource's relations. We now focus onthe “competence realizes requirement”. According torequirement interpretation [5], we interpret this relationas “the disposition of the object satisfies the requirement'sproposition”. We understand that the competence is of thesame disposition type as of the capability that is manifested(or acquired depending on the given interpretation) withthe performance of the behavior element (event). Theinterpretation varies according to the interpretation givento the capability relationship. We label this as “problemC2.1”, since it is a consequence of C2. This can be inter-preted as that (i) by being able of executing a certainbehavior element (an event universal), the resource objecthas a disposition d (of the competence defined type) thatinstantiates the type represented by the capability or that;(ii) s is a situation in which the object has the disposition dand e is an event representing the behavior element, e is apre-state of s, in which d is an instance of the competence.The relation “competence is realized by structure element”relates the dispositions to the elements that bears them.This is interpreted in UFO as that the object that representsthe structure element bears a disposition of that dispositiontype. This is desirable, however, since it is not enforced bythe language, we label the lack of the knowledge on whichobject bears the disposition as “problem C1.1”, since it isassociated to C1.

7.2. Problem C6

The interpretation of the competence concept points tothe same ontological construct as the interpretation of thecapability concept. We label this as “problem C6”. Thecompetence concept appears to have been introduced to fillthe gap in the proposal that it is unknownwhich capabilitiesa resource (or structure element) has. This shows a case ofconstruct redundancy in the language. According to [73],“construct redundancy occurs when more than one gram-matical construct can be used to represent the sameontological construct”. Our analysis confirms and explicatesthe informal suspicions raised in the original proposal textwhen it states that “depending on the (interpretation ofthe) definition of competence, one may argue that, forexample it is more natural to introduce competence inthe metamodel as a specialization of a capability” [41]. Theoriginal proposal also states that “the semantic distancebetween competence, on one hand, and either resource orcapability, on the other hand, is too small”.

8. Well-founded language revision

In this section we propose improvements to the lan-guage based on the ontological analysis, the originalintended meaning for the extension concepts and solutionsto problems revealed in Sections 5–7. We propose a revised

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262246

metamodel to the language extension as part of the solu-tion. We have attempted to preserve the original intendedinterpretations and original relations whenever possible,still addressing the semantic problems. Some additionalexpressiveness is made possible by using relations thatwere not initially employed in the BSVC metamodel, butmost of which already existed in ArchiMate. We refrainfrom defining extensions to the ArchiMate core, focusingonly on the BSVC capability and resource concepts and theirrelations. The competence construct was eliminated as adirect consequence of the ontological analysis.

Fig. 5 presents a fragment of the revised metamodel.The constructs, relations and their semantics are discussedin the sequel.

8.1. Resource

The resource concept in UFO represents a role or a rolemixin that objects may play in particular contexts of usage.In Section 5, problem R1 stated that an element is aresource in a defined context, but it is not a resource inall situations, and that this was not enforced by thelanguage. Thus, a resource should have a defined context.

The metamodel proposed here defines two relationsbetween resources and capabilities, namely one that alreadyexisted in the extension (assigned to), and another due to theArchiMate core (the generic associated with relation).

We propose that the associated with relationshipbetween resource and capability should represent thecontext of usage of a resource. This relation shouldrepresent that a resource is allocated to perform thecapability it is associated to, thus being the performanceof that capability the context of usage of the element thatis playing the resource role. This relation addressesproblem R1.

Another relation involving resources is the assigned torelation, which we use to denote that a resource has thatcapability. It means that the object that instantiates therole represented by the resource has a disposition of thetype represented by the capability. Combining the twoaforementioned relations enables the modeler to repre-sent the prescriptive capabilities that resources shouldhave in order to perform a specific capability (e.g., toperform c1 a resource r should have ca, cb and cc). These

Fig. 5. Fragment of the pr

relations address problems C1, C1.1, C2 and C2.1, partiallycontributing to their solutions.

The metamodel also introduces a controls relationbetween resources and structure elements. The controlsrelation defines that a resource is controlled by an activestructure element. Being controlled is understood as beingavailable for the organization (e.g., by an employmentcontract between employers and employees, or by havingthe right/ownership over a certain object). Since theArchiMate specification defines passive structure elementsas “objects in which behavior is performed”, usuallydenoting information and data elements, we understandthat only active structure elements are to control resources.This relation addresses problem R2.

Problem R6 states that there is no distinction betweenAND or OR resource aggregations in ArchiMate and, assuch, the language lacks the expressivity to state optionalresources. Although ArchiMate does not distinguishbetween AND or OR aggregations, an extension allowingthe representation of optional resources is addressed anddescribed in Section 8.5.

8.2. Structure element and its specializations

The proposed metamodel further introduces the assignedto relation between structure element and capability. Thisrelation defines that a structure element has a capability. Itmeans that the object represented by the structure elementhas a disposition of the type represented by the capability.This relationship enables the organization to specify whichcapabilities it has, as a whole, and as a summary of thecapabilities of its participating parts (organizational units,individual agents, etc.). This relation, in addition to the“resource is assigned to capability relation”, described inSection 8.1, solves problems C1 and C1.1.

The structure element realizes resource relation meansthat a resource is realized/performed by the mentionedstructure element. This enables the organization to be ableto match the “required” capabilities a resource shouldhave, as in its prescriptive version, to the capabilities theorganization's structure elements actually have.

Problem R4 stated that for the cases in which a resourceis realized by a business role (a universal), the languagedoes not determine whether one or more individuals that

oposed metamodel.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 247

instantiate the role (mixin) represented by the resource arerequired in the context of usage. This limitation can beaddressed by the addition of a replication attribute to thelanguage, such as a cardinality constraint. However, thiskind of extension is out of the scope of this work, since itdeals with ArchiMate core. These limitations have beenreported by industry, and some ArchiMate tools alreadyimplement a replication attribute for similar purposes, inwhich case R4 could be considered remediated.

The Structure Element is explicitly shown in the proposedmetamodel with its specializations, active structure elementand passive structure element. Problem R3 states that thelanguage does not distinguish resources that are realized byagents from those realized by non-agentive objects. Theselimitations can be addressed by specializing the structureelement concept to explicit agentive elements and non-agentive elements. The language is also originally unable toaddress the cases in which resources are passive non-agentiveelements, which are objects of interest or raw material (e.g.,“Money”, gold, diamond, gas), which corresponds to problemR5. This can be addressed by specializing the structureelement concept to include these objects of interest and rawmaterial. Both these proposals, however, are out of the scopeof this work, since they deal with the ArchiMate core. TheArchiMate core language or an extension that deals with thecore elements of ArchiMate should address these limitations.

8.3. The behavior element concept

The behavior element realizes capability relation meansthat elements that are capable of performing the behaviorelement have the capabilities that the behavior element issaid to realize. Another reasoning is that some behaviorelements might be seen as the simple ability to performthe specified behavior, as a business process might reflectthe ability to organize elements in a specified manner inorder to produce a desired outcome.

The capability is used by behavior element relation speci-fies that a capability of that type is required to the successfulexecution of the behavior element. In other words, for abehavior element to be performed, one (or more) specificcapability (or capabilities) is required. This allows the orga-nization to structure its necessities in terms of capabilitiesand then to prescribe the necessary capabilities that itsresources should have in order to be able to perform thebehavior element. Further, this scenario would allow theorganization to match the structure elements it has to therequired capabilities for the behavior element it needs toperform, thus improving organizational planning. The plan-ning improves the organizational management of its currentand future structure elements. It also helps the organizationunderstand the elements it needs to acquire according to itsplan. This enables the organization to work aligned to themethods suggested in resource-centric theories.

The behavior element access a capability relation specifiesthat a capability of that type is acquired with the behaviorelement performance. The element that acquires the cap-ability is the one that is assigned to the capability. It meansthat the object assigned to the capability, represented by thestructure element or the structure element that is playingthe resource role, acquires a disposition of the type

represented by the capability by performing the behaviorelement. This relationship allows the organization to specifywhich capabilities its resources and structure elementsacquire and how they acquire that (e.g., through training).This relationship thus solves problem C2 (and as a conse-quence C2.1) in tandem with the following relations:“resource assigned to capability” (Section 8.1), “structureelement assigned to capability” (Section 8.2) and “behaviorelement realizes capability relation”.

8.4. The capability concept

The capability concept is a central concept in thisproposal. The capability concept represents the power tobring about a desired outcome. This power should beunderstood in broad sense, as, for example, a mug hasthe power of constraining coffee, which is the desiredoutcome. Capabilities can be used to state a broad range ofbehaviors, ranging from simple ones as mug's behavior, tocomplex behaviors, as the “damage assessment expertise”(shown in Fig. 2) that can be assumed to inhere in anorganization as well as in a specific person.

Based on the UFO concept of disposition, this interpreta-tion allows the organization to model the capabilities it can“socially perform”. This is applicable for a variety of cases thatuses delegation, such as, for example, cases in which anorganization might hire a different company to bring aboutsome desired outcome and needs to state that the original(hiring) organization has the capability of bringing about thatdesired outcome (in this case because it has the capability ofdelegating it [12,34]). As stated in Section 6, “if A has a meta-commitment from B to execute S then A (socially) can do S. Anobject can have dispositions [capabilities] which arise from itsparts (or from the network of its delegation relations)”.

In the proposed metamodel, the capability concept isrelated to the resource, Structure Element and BehaviorElement concepts, and with itself. The relations with theother elements have been explained in Sections 8.1–8.3,respectively. We now focus on the capability realizes capabilityrelation. Let us assume that C2 realizes C1. This relation meansthat C2 enables C1, in the sense that C2 partially realizes or isrequired for one to have capability C1. As a capability can beenabled by multiple ways and might need different (“lower-level”) capabilities to be enabled, the need for optionalpossibilities arises. The capability-enabling bundle concept isused to address this need.

8.5. The capability-enabling bundle concept

The capability-enabling bundle concept is used to repre-sent the diverse approaches that can be used to enable adesired capability. The bundle is used to represent optionalenabling approaches to a specific capability. This conceptsolves the lack of expressivity stated in problem R6, as wellas the lack of expressivity stated in problem C5. When nobundle is used, we assume a conjunction, i.e., that allelements are required. When more than one bundle thatexplicitly realize the same capability are modeled, eachbundle is considered as an alternative. Fig. 6 presents themetamodel fragment that shows the capability-enabling

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262248

bundle concept and its relations to the other elements ofthe metamodel.

Fig. 7 presents an example of use of the capability andthe capability-enabling bundle. In the example, “Archisur-ance” is assigned to an “Insuring capability”. The “Insuringcapability” is realized by the “Selling Capability”, “thePolicy Administration process”, the “Collect Premium Busi-

Fig. 6. Fragment of the proposed metamodel for the capability-enabling bundle.

Fig. 7. Capability and capability-enabling bundle examples.

ness Function” and the “Claim Handling Capability”, in thesense that they, altogether, are required to enable the“Insuring Capability”. To model alternative means ofenabling a capability, the modeler should use acapability-enabling bundle like the “Claim Handling Cap-ability”, which is realized by “Capability Bundle 1” or by

“Capability Bundle 2”. In Fig. 7, the “Capability Bundle 2”option uses the “Damage Assessment Expert”, and relieson his capabilities to bring about the desired outcome. Incontrast, the “Capability Bundle 1” relies on its “outsour-cing capability” to outsource the required outcome to athird-party.

Problem C3 states that if no resource is represented, the

object manifesting (or acquiring) the capability is unknownat the “capability realized by behavior element” and “resourceassigned to capability” relations. The proposed metamodelrelates resources to capabilities using the associated withrelation for the purpose of indicating the resource thatparticipates on the performance of the capability.

4 This case and its description are taken and adapted from [62].

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 249

Problem C4 states that the language is not expressive tostate the number/amount of resources related to thecapabilities when the resource represents a universal (atype). For example, it is not possible to know how manyinstances of resources are to manifest their capabilities toperform a behavior element. In our proposal, resourcesshould be related to behavior elements through capabil-ities. However, the solution to problem C4 is similar to thesolution to problem R4, described in Section 8.1, so thatthe limitation can be addressed with the addition of areplication attribute to the language, such as a cardinalityconstraint. Nevertheless, this is out of the scope of thiswork since it requires modifications to the ArchiMate core.

9. Applying capability and resource modeling forstrategic management

The extension presented in this paper introduces mod-eling constructs to represent capabilities and resources forhigh-level strategic management, such as described in[6,7,25,48,50,55,58], among others. According to [50],most companies pursue a strategy informally termed as“umbrella strategy”, in which the general strategic guide-lines are initially deliberated, and the details are left to bedeliberated later on in the process.

In this setting, this extension aims to allow organiza-tions to model their core capabilities and resources, bothto understand their current status or to improve theorganization, according to a capability-based planningapproach. The enterprise then focuses on maintaining,improving and creating the required capabilities.

A further key characteristic for the extension is that itenables capabilities and resources to serve as abstractionsfor (more detailed) business process and structure ele-ments. This allows us to define models that should bemore stable and require less effort to maintain thanmodels based on those more detailed elements. Thisallows the enterprise to plan on providing the requiredcapabilities and resources in order to achieve a desiredstate without actually having to pursue a complete andextended view on the business processes and tasks thatare meant to realize that state. The organization mayselectively assign resources to the performance of thecapabilities and/or design new business processes in orderto achieve the modeled desired situation.

In the remaining of this section, we discuss two real-world cases modeled using our proposal. The first case hasbeen taken from the strategic management literature [62]and concerns capability improvement in the automotiveindustry. This case shows the applicability of the capabilityconcept and its relations with other elements of theenterprise architecture. The case is instructive in that itshows that the capability concept directly captures keyaspects of Toyota's “organizational capability enhancement”approach as applied in one its suppliers (JECO). The secondcase concerns portfolio management for an energy suppliercompany. This case models a necessary reconfiguration ofresources in which the enterprise is required to maintainthe same capabilities while consolidating its IT systems.This case has been previously used in the original BusinessStrategy and Valuation Concepts extension proposal [41].

Using this case, we are able to compare our proposal to theoriginal one and present the improvements that the well-founded proposal provides to the language.

9.1. Toyota's JECO supplier use case – organizationalcapability enhancement4

This use case considers supplier development, which isa procedure undertaken by a company to help improve itssupplier's capabilities. This specific case of supplier devel-opment can be understood as an organization's attempt totransfer (or replicate) some aspects of its in-house orga-nizational capability across its boundaries. The organiza-tional capabilities being replicated at suppliers include notonly well-specified technical routines, “but also the rela-tively constant dispositions and strategic heuristics thatshape the approach of a firm to the non-routine problemsit faces” [54]. As stated in [62], “the ability to replicate suchcapability, is, in itself, also a capability”.

We focus on the capability improvement efforts ofToyota in one of its suppliers, the JECO instrumentationsupplier. Specifically, we take the diffusion of the ToyotaProduction System at JECO. For building up the case andmodeling it, we refer to the work published by [62], inwhich three major Japanese automotive industries wereanalyzed. To perform this analysis the authors had accessto historical and contemporary documents provided by thecompanies and also conducted interviews with keyrespondents (purchasing managers and supplier develop-ment engineers) in the organization and at some of itssuppliers.

The Toyota Production System (TPS) has its roots inthe Toyota factories of the 1960s. The Operations Manage-ment Consulting Division (OMCD) was established as partof Toyota's Production Control Function, to facilitate aseamless transfer of knowledge between Toyota and itssuppliers. OMCD is in charge of implementing TPS bothwithin Toyota factories and its core suppliers, and itintends to guarantee that the same methods, proceduresand heuristics are applied to both internal and externalfactories.

Within Toyota, kojo jishuken (factory autonomous studygroups) take place as a culmination of education andtraining for Toyota's middle managers and first-line super-visors. They are considered the most important repositoryof know-how on the shop floor. Supervisors are given anincentive to make continuous improvement with concreteresults, and are required to regularly present improvementideas. The transfer of knowledge and reproduction ofToyota's capabilities using jishuken groups (self studyautonomous groups) in its suppliers did not become publicuntil the 90s [62].

Fig. 8 shows Toyota's intentions, represented by itsgoals, drivers and assessments, represented using Archi-Mate's motivation extension notation. In our case, Toyotahas the goals (i) “Provide Supplier Development”; (ii)“Human Resource Development”; (iii) “Supplier Able toAdapt to Demand Fluctuations”; (iv) “Supplier Able to

Fig. 8. Toyota's intentions.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262250

Adapt to Model Mix Changes”; (v) “Transfer of Knowledgebetween Toyota and its Suppliers”; (vi) “Reduce Depen-dency of Specific Supplier”; (vii) “Engage in Target Cost-ing”; and specific goals for the application of TPS in JECO,(viii) “Reduce Inventory by 54% (to meet orders fluctua-tions)” and (ix) “Reduce The Inventory of Rotors by 95%”.Toyota had some drivers for some of its goals, like the“Demand Fluctuations” and the “Model Mix Changes”drivers.

The performance of one goal might influence the othergoals. In our example, goals viii and ix, influences goals i, iiiand/or iv. An important assessment that has been made byToyota was that “supplier has no just in time production”.The crisis in the automotive sector, in which many ofToyota's suppliers were experiencing difficulties adaptingto Demand Fluctuations, enforced this concern.

Fig. 9 shows an ArchiMate model that represents anoverview of Toyota's capability improvement efforts forJECO, including Toyota's capabilities, JECO's capabilities,the capabilities of the so-called autonomous study groups(“Jishuken study group”), and the key resources.

The “Jishuken study groups” were designed to helpsuppliers improve their shop floors by refining the appli-cation of TPS. The “Jishuken study group” is required tochoose a specific theme amongst the ones discussed by theOMCD, and to identify a specific factory area to be studiedby the group. JECO study group chose the output fluctua-tions in the age of low demand (“Demand Fluctuations”)and the “Model Mix Changes”. The specific factory areachosen for the study was “JECO Parts and Goods logistics”.Toyota has considered mainly geographical location andthe absence of direct competitors when forming thesegroups, to help interaction and sharing of know-howduring the “jishuken activities”.

The study section begins by setting concrete perfor-mance targets in terms of shop floor indicators, whichshould be achieved with the help of the “Jishuken activ-ities”. Typical indicators to be considered are productivity,cost reduction and/or inventory turns. JECO's targets wereto “Reduce Inventory by 54%” and to “reduce the inventoryof rotors by 95%”. This is represented in the model by the

dotted lines with a white arrow on the end between theactivity and the goal it realizes.

From the Toyota side, trained experts (“Supplier Devel-opment Employees”) help a jishuken group to achieve itsresults. A “Senior OMCD engineer” with “observation andanalysis capability” is in charge of the “jishuken studygroup” and visits the company under study duringthe “Jishuken Activity” period, in which he is respon-sible for making critical observations. “Junior SupplierDevelopment Engineers” also visit the company, moreoften though, to give more detailed guidance.

Jishuken groups are responsible to put forward and“Implement Concrete Improvement Ideas”. A typical Jishu-ken group would consist of engineers from various suppli-ers, not only for the company under study, since thisfacilitates knowledge sharing, which is important for the“Developing and Training People Capability” realized bythe “Jishuken Activity”. Most of the ideas are implementedin the study-host company. For example, JECO has imple-mented 222 of the 248 improvement ideas, an implemen-tation rate of 90%. Many of the ideas implemented by JECOconcerned the improvement of its “Kanban (Just in TimeScheduling) Capability” and its “Handling Defects Capabil-ity”, with the clarification of rules about when defects arediscovered. These capabilities helped JECO accomplish the“Just In Time Manufacturing Capability”, a specialization ofits “Manufacturing Capability”, which realizes its “Just inTime Delivery of Parts Capability”, used in its “Just In TimeDelivery of Parts” service. The “Manufacturing Capability”by itself would allow the enterprise to realize the “Just InTime Delivery of Parts” service, since the organizationmight have all the required parts in its inventory. However,the “Just In Time Manufacturing” capability has beenenable by the “Kanban (Just in time Scheduling) Capabil-ity” and the “Handling Defect Capability”. This helped theenterprise to diminish its inventory of rotors and itsregular inventory, realizing the “Just in Time Manufactur-ing Capability” and, thus, addressing Toyota's assessmentthat “Supplier has no just in time production”, and realizingJECO's goals of “reducing inventory by 54% (to meet orderfluctuations)” and “Reduce The Inventory of Rotors by

capab

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 251

95%”. These goal realizations, in their turn, influence therealization of Toyota's goals of “Provide Supplier Develop-ment”, “Supplier Able to Adapt to Model Mix Changes” and“Supplier Able to Adapt to Demand Fluctuations”.

Besides Jishuken, Toyota's OMCD also provides indivi-dual assistance to suppliers on an if-and-when-necessarybasis. For that, OMCD have the capability of “Deep SupplierIntervention” and “Individual Supplier Assistance”. Indivi-dual assistance is suitable whenever Toyota is looking for

Fig. 9. Toyota's supplier

quick results. However, short-term deep intervention issaid to come to a halt when the experts go home, since thesuppliers do not understand why things are being per-formed as intervened and short-term deep interventiondoes not “provide supplier development” or “humanresource development”. A Toyota OMCD expert under-stands that the “[i]t would most certain be quicker for anexpert to take a lead and provide answers, but this wouldnot result in developing the skills of those who are led. The

ility enhancement case.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262252

strength of Toyota Production System lies in creating asmany people who can implement and put into practiceTPS on their own as possible. So the most important thingfor the survival of TPS is ‘Human Resource Development’.”

Jishuken is a gathering of middle-level productiontechnologists from a stable group of companies, whojointly develop better capabilities for applying TPS throughmutual criticism and concrete application. Jishuken alsohas the benefit of giving Toyota “Supplier Costs Knowl-edge”, with an enormous access to detailed cost structuresof its main suppliers, a capability that inheres in theOMCD. This contributes to Toyota's core capability of“Engaging in Target Costing” and to retention of “Manu-facturing know-how for components not produced in-house”.

In order to evidence the role of the capability construct,a model without the capability construct is presented inFig. 10. This model is instructive in that it shows thatwithout the capability construct the model is not able tocapture key aspects of Toyota's organizational capabilityenhancement approach. Instead, it focuses on more opera-tional aspects of the same enterprise setting, failing toreveal the capabilities of JECO that are at stake, the keycapabilities of Toyota's OMCD, as well as the link betweenthe activities of the study group and the capabilitiesimproved.

9.2. IT consolidation for an European Energy SupplierCompany

The second case we use to illustrate our approachconcerns an IT consolidation problem [41]. Consolidationof software application portfolios is a typical situation inwhich portfolio management techniques are applied in acapability-based planning setting. The main goals of theconsolidation of IT resources are the elimination of func-tional and data redundancies. Typical situations in whichIT consolidation is necessary include the co-existence ofdifferent software systems in an organization that offer thesame functionality, or the replication and storage of databy several different systems. The positive effect of ITconsolidation on cost reduction has long been recognizedin the literature, such as, e.g., in [35]). In particular, werefer to [18] that used integer binary programming to solvethis IT consolidation problem, while minimizing consoli-dation and maintenance costs. In this particular case, theorganization seeks IT consolidation to gain efficiency with-out affecting its business capabilities. Expressing businesscapabilities is thus required in order to make their coher-ent management possible and plan in a way that matchesthe organization's business model and strategy. The strat-egy in this case involves gaining efficiency throughresource optimization, while the model requires the main-tenance of the current capabilities.

The goal of this case is twofold:

�

Firstly we demonstrate the benefits of the proposedimproved metamodel fragment concerning the qualityof models compliant to it. We do this by showing howsome of the problems concerning the original

metamodel are handled in a concrete setting with therevised construct and relations.

�

Secondly, we demonstrate its usefulness in making therelation between business strategy and enterprise archi-tecture explicit, allowing one to reason about it. We showhow capability-based planning bridges the gap betweenstrategy definition and strategy implementation.

The capability-based planning process we followthroughout the case consists of the following steps:

1.

Identifying drivers and problem(s), business strategyand strategic capabilities;

2.

Analyzing the baseline capability gaps and creatingbaseline heat capability map;

3.

Relating capability gaps to the baseline architecture; 4. Using resource portfolio management approaches to create

a target architecture and target capability map, and;

5. Implementing changes.

We consider and extend the case that was reported in[41], namely of a large European Energy Supplier (EES)that consists of three different units (formerly threedifferent companies that have merged). As a result of theenergy market liberalization, customers may now switchbetween energy suppliers, and thus this energy suppliermust be able to ensure a fast and reliable switchingprocess for new and leaving customers. A consequence ofthe previous merger is that currently the company hasseven different systems that all take care of some part ofcustomer switching process for three business units, andprovide overlapping functionality. Information about cus-tomers, contracts and their consumptions is scattered overdifferent systems and databases. The company aims atconsolidating the system architecture, and at eliminatingredundant functionality with minimal costs. We presentthe results of each step of the capability-based planningprocess in the sequel, employing ArchiMate models, someof which include the capability and resource constructs.We omit discussing step 5 (implementing the changes) asthis is outside the scope of this paper.

9.2.1. Identifying drivers and problem(s), business strategyand strategic capabilities

In this case, the main driver for action was the energymarket liberalization. An assessment of EES's current situationin the light of this driver is shown in Fig. 11 using ArchiMate'smotivation extension. It relates the driver (“Energy marketliberalization”) and various assessments (“High IT maintenancecosts”, “Inefficient switching process”, etc.).

The assessments are related to many of the goals thatfall under the overarching “Achieving operational excel-lence” business strategy (Fig. 12).

9.2.2. Analyzing baseline capability gaps and creatingbaseline heat capability map

Fig. 13 shows the relation between the assessments andthe capabilities whose current realization leads to the

Fig. 10. Toyota's case modeled without the capability concept.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 253

challenges identified in the assessments. The identifiedcapabilities are “Customer order management”, “Customerdata management”, “Customer relation management” and“Risk management”. The lower part of Fig. 13 shows anumber of operational capabilities that are related to thestrategic capabilities. Figs. 12 and 13 show actually the

upper and lower part of a single diagram, enabling us totrace from strategic elements to capabilities.

The heat map shown in Fig. 14 shows the identifiedstrategic capabilities in context. Strategic capabilitiesdepicted in red are those whose realization needs to beimproved in the context of the addressed problem.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262254

9.2.3. Relating capability gaps to the baseline architectureHaving identified the strategic capabilities gaps, the

next step in the approach concerns relating the capabilitygaps to the baseline architecture, in order to find paths forpossible improvement in capability implementation.

The baseline architecture that focuses on the switchingprocess is shown in Fig. 15. It includes seven differentapplication systems that all take care of switching for threebusiness units, and thus provide overlapping functionality.

The main goal of this step is to establish a relationbetween the previously identified capability gaps andthose elements/parts of the baseline architecture thatshould be changed to resolve the capability gaps, andsubsequently to solve the original problem. Nevertheless,examining the heat-map (Fig. 14) and the baseline archi-

Fig. 12. Relating assessm

Fig. 11. Assessmen

tecture (Fig. 15) one can see that the distance between thetwo models in terms of abstraction level is large, making itunfeasible to establish a precise relationship betweenthem as far as the traceability of capability gaps in theenterprise architecture is concerned.

In order to address this issue, we add an intermediaryabstraction level between the baseline architecture andthe heat map. This additional level focuses on the speci-fication and refinement of capabilities and resources, sothat traceability all the way down to the operationalaspects of the baseline architecture becomes possible. Inthis process we make use of the links between thestrategic capabilities and the operational capabilitiesshown in the bottom part of Fig. 13. This refinement stepis important because it allows operational capabilities to

ents and strategy.

t and driver.

Fig. 13. Strategic input for capability-based planning: assessment of capability gaps.

Fig. 14. Heat-map of strategic capabilities associated with the “Achieving operational excellence” strategy in the context of the current case. (Forinterpretation of the references to color in this figure, the reader is referred to the web version of this article.)

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 255

be easily linked to architecture fragments. Thus, Fig. 16depicts the configuration of resources and capabilities thatare involved in the current decision making problem, andrepresents their usage context in a simple way.

In Fig. 17 we show how the previously identifiedconfiguration of resources and capabilities is mapped ontothe architecture by focusing on the IT components thesecapabilities require.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262256

9.2.4. Using resource portfolio management approaches andcreating a target architecture and target capability map

As explained in [41], the model shown in Fig. 17 canbe used as basis for the application (in a model-basedfashion) of integer binary programming [18] to enterprisearchitecture once it has been enhanced with the conceptscovered by the proposed metamodel. The resource portfoliooptimization problem to be solved can be shortly formulatedas follows: the objective function is to minimize the switch-ing costs, which are defined as the sum of maintenance andconsolidation costs, subject to two constraints: (1) all pro-cesses must remain functional at all times, meaning that theBus' switching processes must be all fully supported by thesystems that will be selected in the future situation, and (2)no functionality loss is acceptable, meaning that the selectedsystemsmust provide all the functionality currently providedby all of them. Details on how maintenance and consolida-tion costs are calculated can be found in [18,41].

The optimal solution provided by the solver for the costminimization problem leads to the target switching archi-tecture shown in Fig. 18, which includes resources, cap-abilities and relations defined according to our proposed

Fig. 15. Baseline a

metamodel. Systems S2, S4, S5 and S7 have been removedfrom the target architecture, and the remaining systemshave to be enhanced with new process interfaces (coloredin red in Fig. 18) to enable the execution of all BUs'switching processes in the future situation.

9.2.5. Concluding remarksThe case demonstrates the usefulness of the resource

and capability concepts to support a capability-basedplanning process, which is typically carried out at astrategic/tactical level. For such endeavors one should beable to abstract from the details of architecture models butstill being able to trace planning decisions concerningresources and capabilities in these architecture models inorder to correctly assess the impact they have on thedifferent layers of the architecture. In this particular casewe have shown how the problem of multiple, partlyoverlapping, and hardly maintainable distributed ITresources (Fig. 17) can be solved by consolidating theminto a single integrated resource (Fig. 18), which can becontrolled at a corporate level.

rchitecture.

Fig. 16. Resources and capabilities configuration.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 257

Fig. 18 shows some relations that could not be repre-sented with the original metamodel, which are drawn withthick lines. These relations facilitate the correct modelingand interpretation of different dependencies betweenresources, capabilities and architecture elements. In particu-lar, the model shows the relation between the consolidatedresources and the controlling structural element, the relationbetween the resource and its context of usage, and finally,the relation between the capabilities and the (customer-facing) processes that employ the capabilities. Together withthe other models discussed here, traceability from businessstrategy, business capabilities, and operational capabilities to(IT) resources has been made possible.

10. Related work

Some related approaches have addressed the use ofcapabilities and resources in enterprise architecture andenterprise modeling.

The REA accounting model [47], e.g., has included anotion of resource since its inception. In REA, Resourcesare exchanged through economic Events in which Agentsparticipate, hence the REA acronym. Since REA was con-ceived as a conceptual framework to support accountingpractices, its main focus is on the notion of economicresource as a transactable entity of value. It considers,among other aspects, the “flow” of resources and theaccountability for the “custody” of resources (in the“stock” or “inventory”). Similarly to our approach, the

authors of REA have been concerned with the ontologicalfoundations of their framework, which has been reportedin [21,22]. Differently from our work, REA does not aim atconsidering the properties or capabilities of resources, andit does not account for how resources and their capabilitiesmay be employed in the organization in order to achievehigher-level capabilities.

The notion of resource is also adopted in the ARISframework [65], in order to account for resource bottle-necks and resource availability in the scope of businessprocess management. ARIS captures the relation betweenresources and organizational entities that controls them, aswell as “input” and “output” relations between resourcesand events (“functions” in ARIS). The ARIS framework doesnot address the notion of capability.

Dryer et al. [16] discuss the extension of DoDAF with aCapability Evaluation Model that include the concept ofcapability. In that work, capabilities are said to be providedby “systems of systems”, which comprehend any combina-tion of “doctrine, organization, training, materiel, leadershipand education, personnel and facilities (DOTMLPF)”. Some ofthese elements subsumed under DOTMLPF are incorporatedinto the DoDAF's capability viewpoints, DoDAF's DM2 modeland grounded in the IDEAS foundational ontology, which isused as a foundation for the whole DM2. Further investiga-tion is required in order to assess the relations betweenArchiMate BSCV constructs and DoDAF Capability View-points, and to establish potential semantic interoperabilityrelations for the representation of capabilities and resources.

Fig. 17. Baseline architecture with capabilities and resources.

Fig. 18. Target consolidated architecture. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262258

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 259

This task can be made more manageable now that the BSCVconstructs have been given a precise semantics in terms of afoundational ontology. That should enable comparison andharmonization of both frameworks.

In [9], TOGAF has been extended to support themodeling of the capabilities a Business Component (BC)can perform. A BC is a business unit that encompasses a setof activities, supported by assets including people, pro-cesses and technology. The approach uses capabilities as“an idealized conceptual structure that describes what aBC can do to create value for customers”.

In recent years, a number of enterprise modelingapproaches have been subject to ontology-based analysis.In [59], the authors performed an ontological analysis to theBusiness Process Modeling Notation (BPMN). Nine ontologi-cal deficiencies related to modeling when using the BPMNwere found. In [64], the authors have defined the semanticsof the ARIS framework concepts and relationships in terms ofUFO. Problems regarding the ARIS Method were exposed,and possible solutions to these problems were proposed. [5]performed an ontological analysis of the ArchiMate motiva-tion extension proposal, unveiling problems and proposingimprovement recommendations. To the best of our knowl-edge there are no comparable analyses addressing themodeling of capabilities and resources.

11. Conclusions and future work

In this paper, we have discussed an ontological analysis ofthe BSVC ArchiMate extension and the associated notions ofcapability, resources and competences. We have employeda comprehensive foundational ontology that incorporatesconcepts to deal with objects, relations, roles, events, disposi-tions, as well as social and intentional concepts. Our main aimhas been to clarify the semantics of the proposed modelingconstructs, which should contribute to the application of thelanguage in practice as a communication tool for stakeholdersinvolved in decision making. We have revised the originallanguage metamodel, in order to accommodate the proposedrecommendations.

We have been able to clarify that the resource elementrepresents a type-level entity, capturing the role of an(agentive or non-agentive) object in a particular context ofusage. Our well-founded recommendations should lead toa language that allows improved resource planning, as itallows the specification of the context of usage ofresources and the ability to explicitly state mandatoryand optional resources when related to capabilities.

By considering capabilities as a type of dispositions, wehave been able to account for what it means for a behaviorelement to realize a capability. Our recommendationsshould lead to a language that allows improved resourceand capability oversight, as it allows one to relate cap-abilities and structure elements that possess capabilities,as well as relate resources with capabilities.

By also considering that competences should be inter-preted as types of capabilities, we have been able toidentify a case of construct redundancy. We have tracedthe root of the problem to a lack of relations to express thecapabilities of resources. Our recommendations seem tolead to a more regular and parsimonious solution for the

expression of human resources and their capabilities,which was not fully addressed with the introduction ofthe competence construct. Our extension is intended tomodel the organization's core capabilities and keyresources with a strategic management focus. With ourextension, the enterprise can consider the required cap-abilities and resources to achieve a desired state withoutactually having to pursue a complete and extended viewon the business processes and tasks that are necessary torealize that state. This means that at the strategic manage-ment level enterprise architects are able to focus on theproper level of abstraction, avoiding unnecessary commit-ments with lower level details. Thus, they are able toaccommodate future changes in the operational parts ofthe enterprise architecture and at the same time realizethe higher-level capabilities and resources. This creates aloose coupling between higher-level capabilities and otheroperational enterprise architecture elements, contributingto flexibility and maintainability of the resulting enterprisearchitecture descriptions. This approach is especially valu-able to competitive and changing environments, whichrequires both planning and ability to adapt.

The contributions in this paper can be assessed fromthe perspectives of rigor and relevance (as proposed fordesign-science research by [39]). From the perspective ofrigor, we have adopted a well-established ontologicalanalysis methodology (Section 3) and clearly definedontological foundations (Section 4). From the perspectiveof relevance, there is ample support from the businessliterature discussing the key role of capabilities andresources in strategic management [6,7,37,58]. Further,capability-based planning has had major interest fromresearch efforts in the literature as well as from thepractice in EA, with frameworks such as TOGAF introdu-cing basic notions of capability-based planning and its rolein designing, planning and implementing organizationalchange [70]. Incorporating capability-based planning in EAis thus relevant for industrial and academic efforts.

In order to further stress the relevance of the approach,we have shown that it is able to represent key aspects oftwo real-world cases, one concerning capability improve-ment and the other concerning resource consolidation incapability-based strategic management. In the first casestudy, we show that without the capability construct theEA model is not able to capture key aspects of Toyota'sorganizational capability enhancement approach. The EAmodel without the capability construct is focused mostlyon operational aspects loosing the link to “what is rele-vant” for strategic management. In other words, the modelshows operational elements without revealing the keycapabilities that enable achieving enterprise's goals. Inthe second case study, we show the application of theapproach in portfolio management for an energy suppliercompany. More specifically, we show the usefulness of thecapability-based models in representing a real scenariothat required the reconfiguration of resources when theenterprise faced major changes. In this case, the enterprisewas required to maintain the same capabilities whileconsolidating its IT systems. The model captures capabil-ities that remained stable even in the face of changes inthe operational elements of the architecture. It also

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262260

captures traceability from business strategy, business cap-abilities and operational capabilities to (IT) resources. Thiscase study has also served for us to compare our proposalto the original “Business Strategy and Valuation Concepts”extension [41] and present the improvements that thewell-founded proposal provided to the language.

In our future efforts, amongst others we intend toinvestigate the relation between the concepts of resourceand value. The latter deserves careful attention from theperspective of semantic definition and ontological analysis,given its subjective nature and usage flexibility.

Similarly to the discussion in [5] concerning the analysisof the motivation extension to ArchiMate, we do not intendto suggest that the terminology used in this paper shouldreplace the terminology currently used in ArchiMate, and wedo not intend to imply that the UFO conceptualizationshould be exposed directly to users of the standard. Themain role of the ontological analysis has been to provide uswith a rigorous framework to analyze the BSVC proposal. Inthis sense, the ontological analysis can be seen as a tool forhypothesis formulation, and the recommendations that wehave identified here using ontological analysis should beconsidered as subject to further examination. For example,considering the pragmatic impact of amendments on the setof standards and its users. We have outlined the recommen-dations raised by the ontological analysis performed here,and we believe that they can have direct application in therevision of the proposal before it reaches the standardizationeffort. It is natural that, once the approach is incorporatedinto standards and into EA tools, new insights will arise fromthe appropriation of the technique by the industry. Theseinsights should inform an ex post evaluation effort.

Acknowledgments

This research is funded by CNPq (485368/2013-7,311313/2014-0, 461777/2014-2 and 201495/2014-7),CAPES (402991/2012-5) and FAPES (59971509/12).

References

[1] Joao Paulo A. Almeida, Giancarlo Guizzardi, Santos Paulo Sergio Jr.,Applying and extending a semantic foundation for role-related con-cepts in enterprise modelling, Enterp. Inf. Syst. 3 (2009) 253–277⟨http://www.tandfonline.com/doi/abs/10.1080/17517570903046292⟩.

[2] João Paulo A. Almeida, Giancarlo Guizzardi, An ontological analysisof the notion of community in the RM-ODP enterprise language,Comput. Stand. Interfaces 35 (3) (2013) 257–268, http://dx.doi.org/10.1016/j.csi.2012.01.007 ⟨http://linkinghub.elsevier.com/retrieve/pii/S0920548912000402⟩.

[3] Pino G. Audia, Edwin A. Locke, Ken G. Smith, The paradox of success:an archival and a laboratory study of strategic persistence followingradical environmental change, Acad. Manag. J. 43 (5) (2000)837–853.

[4] Carlos L. B. Azevedo, Maria-Eugenia Iacob, João Paulo A. Almeida,Marten van Sinderen, Luís Ferreira Pires, Giancarlo Guizzardi, 2013,An ontology-based well-founded proposal for modeling resourcesand capabilities in ArchiMate, in: Proceedings of the 17th IEEEInternational Enterprise Distributed Object Computing Conference(EDOC), pp. 39–48.

[5] Carlos L. B. Azevedo, Joao Paulo A. Almeida, Marten van Sinderen,Dick A.C. Quartel, Giancarlo Guizzardi, 2011, An ontology-basedsemantics for the motivation extension to ArchiMate, in: Proceed-ings of the 2011 IEEE 15th International Enterprise DistributedObject Computing Conference. doi:10.1109/EDOC.2011.29. ⟨http://

ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6037557⟩.

[6] Tim S. Baines, Howard W. Lightfoot, Ornella Benedettini, JohnM. Kay, The servitization of manufacturing: a review of literatureand reflection on future challenges, J. Manuf. Technol. Manag. 20 (5)(2009) 547–567, http://dx.doi.org/10.1108/17410380910960984⟨http://www.emeraldinsight.com/10.1108/17410380910960984⟩.

[7] Jay Barney, Firm resources and sustained competitive advantage, J.Manag. 17 (1) (1991) 99–120.

[8] I. Barreto, Dynamic capabilities: a review of past research and anagenda for the future, J. Manag. 36 (1) (2009) 256–280, http://dx.doi.org/10.1177/0149206309350776.

[9] Tiago Barroero, Gianmario Motta, Giovanni Pignatelli, 2010, Businesscapabilities centric enterprise architecture, in: Enterprise Architec-ture, Integration and Interoperability, Springer, pp. 32–43.

[10] Conrad Bock, James Odell, A more complete model of relations and theirimplementation: roles, J. Object Oriented Program. 11 (1998) 51–54.

[11] François Bodart, Arvind Patel, Marc Sim, Ron Weber, Should optionalproperties be used in conceptual modelling? A theory and threeempirical tests, Inf. Syst. Res. 12 (4) (2001) 384–405.

[12] Ana C. O. Bringuente, Ricardo A. Falbo, Giancarlo Guizzardi, 2010,Using a Foundational Ontology for Reengineering a Software ProcessOntology.

[13] Evellin Cristine Souza, Cardoso, Paulo Sérgio Santos Jr., PauloAndrade Almeida João, Renata S.S. Guizzardi, Giancarlo Guizzardi,2010, Semantic integration of goal and business process modeling,in: Proceedings of the International Conference on Research andPractical Issues of Enterprise Information Systems (CONFENIS 2010).

[14] Mohammad Hossein, Danesh, Yu Eric, 2014, Modeling EnterpriseCapabilities with I* , pp. 112–123.

[15] Islay G. Davies, Michael Rosemann, Peter Green, 2000, Exploringproposed ontological issues of ARIS with different categories ofmodellers, in: Proceedings of the Australasian Conference on Infor-mation Systems, Hobart, Australia.

[16] David A. Dryer, Thomas Bock, Maria Broschi, Timothy D. Beach, 2007,DoDAF limitations and enhancements for the capability test meth-odology, in: Proceedings of the 2007 Spring Simulation Multicon-ference, SpringSim '07, Volume 3, Society for Computer SimulationInternational, San Diego, CA, USA, pp. 170–176. ⟨http://dl.acm.org/citation.cfm?id=1404803.1404839⟩.

[17] Kathleen M. Eisenhardt, Jeffrey A. Martin, Dynamic capabilities:what are they? Strateg. Manag. J. 21 (10–11) (2000) 1105–1121.10.1002/1097-0266(200010/11)21:10/11o1105::AID-SMJ13343.0.CO;2-E. http://doi.wiley.com/10.1002/1097-0266(200010/11)21:10/11o1105::AID-SMJ13343.0.CO;2-E..

[18] Ulrik Franke, Oliver Holschke, Markus Buschle, Per Närman, JannisRake-Revelant, 2010, IT consolidation—an optimization approach, in:Proceedings of the IEEE International Enterprise Distributed ObjectComputing Workshop, EDOC, pp. 21–26. doi:10.1109/EDOCW.2010.11.

[19] Frederik Gailly, Guido Geerts, Geert Poels, 2009, Ontological reengineer-ing of the REA-EO using UFO, in: Proceedings of the InternationalOOPSLA Workshop on Ontology-Driven Software Engineering.

[20] Aldo Gangemi, Nicola Guarino, Claudio Masolo, Alessandro Oltramari,Luc Schneider, Sweetening ontologies with DOLCE in knowledge engi-neering and knowledge management: ontologies and the semantic Web,Lect. Notes Comput. Sci. 2473 (2002) 223–233, http://dx.doi.org/10.1007/3-540-45810-7_18. http://dx.doi.org/10.1007/3-540-45810-7_18.

[21] Guido L. Geerts, Wiliam E. McCarthy, 2000, The ontological founda-tion of REA enterprise information systems, in: Proceedings of theAnnual Meeting of the American Accounting Association, pp. 1–34.⟨http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.91.8191&rep=rep1&type=pdf⟩.

[22] Guido L. Geerts, William E. McCarthy, An ontological analysis of theeconomic primitives of the extended-REA enterprise informationarchitecture, Int. J. Account. Inf. Syst. 3 (2002) 1–16, http://dx.doi.org/10.1016/S1467-0895(01)00020-3.

[23] Andrew Gemino, Yair Wand, Complexity and clarity in conceptualmodeling: comparison of mandatory and optional properties, DataKnowl. Eng. 55 (3) (2005) 301–326.

[24] Jaap Gordijn, Hans Akkermans, Designing and evaluating E-businessmodels, IEEE Intell. Syst. 16 (4) (2001) 11–17, http://dx.doi.org/10.1109/5254.941353. http://dx.doi.org/10.1109/5254.941353.

[25] Robert M. Grant, Toward a knowledge-based theory of the firm,Strateg. Manag. J. 17 (1996) 109–122.

[26] Peter Green, Michael Rosemann, 2005, Ontological analysis ofbusiness systems analysis techniques: experiences and proposalsfor an enhanced methodology, in: Business Systems Analysis withOntologies, ICI Global, pp. 1–27.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262 261

[27] Nicola Guarino, 1998, Formal Ontology and Information Systems,pp. 3–15.

[28] Giancarlo Guizzardi, Ontological Foundations for Structural Concep-tual Models, Univeristy of Twente, 2005.

[29] Giancarlo Guizzardi, Ricardo Falbo, Renata S.S. Guizzardi, 2008,Grounding software domain ontologies in the unified foundationalontology (UFO): the case of the Ode software process ontology, in:Proceedings of the XI Iberoamerican Workshop on RequirementsEngineering and Software Environments, pp. 244–251.

[30] Giancarlo Guizzardi, Gerd Wagner, 2008, What's in a relationship:an ontological analysis, in: Conceptual Modeling-ER 2008, Springer,pp. 83–97.

[31] Giancarlo Guizzardi, Gerd Wagner, Ricardo de Almeida Falbo, RenataS.S. Guizzardi, João Paulo A. Almeida, 2013, Towards ontologicalfoundations for the conceptual modeling of events, in: ConceptualModeling, Springer, 327–341.

[32] Giancarlo Guizzardi, Gerd Wagner, Nicola Guarino, Marten vanSinderen, 2004, An ontologically well-founded profile for UMLconceptual models, in: Advanced Information Systems Engineering,Springer, pp. 112–126.

[33] Renata S.S. Guizzardi, Giancarlo Guizzardi, Anna Perini, John Mylo-poulos, 2007, Towards an ontological account of agent-orientedgoals, in: Software Engineering for Multi-Agent Systems V, Springer,pp. 148–164.

[34] Renata S.S. Guizzardi, Giancarlo Guizzardi, 2011, Ontology-based trans-formation framework from TROPOS to AORML, in: Social Modeling forRequirements Engineering, MIT Press, Cooperative Information SystemsSeries, pp. 547–570. ⟨http://ufesmestradoanabringuente.googlecode.com/svn-history/r278/trunk/4per/Artigos/CE9917E7d01.pdf⟩.

[35] Diana Hancock, David B. Humphrey, James A. Wilcox, Costreductions in electronic payments: the roles of consolidation,economies of scale, and technical change, J. Bank. Financ. 23 (2)(1999) 384–405, http://dx.doi.org/10.1016/S0378-4266(98)00108-3.

[36] John Heil, From an Ontological Point of View, Clarendon Press,Oxford, 2003.

[37] Constance E. Helfat, E. Constance, G. Sidney, S.G. Winter, Untanglingdynamic and operational capabilities: strategy for the (N) ever-changing world, Strateg. Manag. J. 1250 (2011) 1243–1250, http://dx.doi.org/10.1002/smj ⟨http://onlinelibrary.wiley.com/doi/10.1002/smj.955/full⟩.

[38] Heinrich Herre, Barbara Heller, Patryk Burek, Robert Hoehndorf,Frank Loebe, Hannes Michalek, General formal ontology (GFO)—afoundational ontology integrating objects and processes, Onto-MedRep. (2006) 8.

[39] Alan R. Hevner, S.T. March, J. Park, S. Ram, Design science in informationsystems research, MIS Q. 28 (1) (2004) 75–105, http://dx.doi.org/10.1007/s12599-012-0246-0 ⟨http://dl.acm.org/citation.cfm?id=2017217⟩.

[40] Maria-Eugenia Iacob, H. Jonkers, M.M. Lankhorst, E. Proper, D.A.C.Quartel, 2012, ArchiMate 2.0 Specification: The Open Group.

[41] Maria-Engenia Iacob, Dick A.C. Quartel, Henk Jonkers, 2012, Captur-ing business strategy and value in enterprise architecture to supportportfolio valuation, in: Proceedings of the 2012 IEEE 16th Interna-tional Enterprise Distributed Object Computing Conference (Sep-tember), pp. 11–20. doi:10.1109/EDOC.2012.12. ⟨http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6337268⟩.

[42] Jeroen Kraaijenbrink, J.-C. Spender, Aard J. Groen, The resource-based view: a review and assessment of its critiques, J. Manag. 36 (1)(2010) 349–372.

[43] Wim Laurier, Geert Poels, 2009, Enterprise Ontology-Based Struc-turing of Conceptual Data Modeling Patterns.

[44] Wim Laurier, Geert Poels, 2009, Extending REA models with areference model for abstraction mechanisms, in: Proceedings ofthe 4th International Workshop on Value Modeling and BusinessOntologies, Amsterdam.

[45] Hyeoncheol Lee, Yeong T. Song, 2011, Bridging enterprise architec-ture requirements to Archimate, in: Studies in ComputationalIntelligence, Springer, 365, pp. 63–78. 10.1007/978-3-642-21375-5_6.

[46] E. Jonathan Lowe, The Four-Category Ontology: A MetaphysicalFoundation for Natural Science, Clarendon Press, 2006.

[47] William E.W.E. McCarthy, The REA accounting model: a generalizedframework for accounting systems in a shared data environmentdoi:8, Account. Rev. 57 (1982) 554–578.

[48] Alexander McKelvie, Per Davidsson, From resource base to dynamiccapabilities: an investigation of new firms http://doi.wiley.com/10.1111/j.1467-8551.2008.00613.x, Br. J. Manag. 20 (2009) S63–S80,http://dx.doi.org/10.1111/j.1467-8551.2008.00613.x.

[49] Josef Miklos, A meta-model for the spatial capability architecture, J.Theor. Appl. Inf. Technol. 43 (2012) 2.

[50] Henry Mintzberg, The rise and fall of strategic planning, Harv. Bus. Rev.72 (1994) 107–114, http://dx.doi.org/10.1016/0024-6301(94)90173-2⟨http://www.hotlinkfiles.com/files/996068_cfudn/Strategic_Planning-Mintzberg.pdf⟩.

[51] George Molnar, in: Mumford Stephen (Ed.), Powers: A Study inMetaphysics, Oxford University Press, 2006.

[52] J Scott, Business Capability Maps – The missing link betweenbusiness strategy and IT action., Architecture & Governance 5 (9)(2009) 1–4.

[53] Stephen Mumford, Dispositions, Oxford University Press, 2003.[54] Richard R. Nelson, Sidney G. Winter, An evolutionary theory of

economic change, Econom. J. 93 (1982), http://dx.doi.org/10.2307/2232409 ⟨http://books.google.com/books?id=6Kx7s_HXxrkC&pgis=1⟩.

[55] Edith Tilton Penrose,, The Theory of the Growth of the Firm, OxfordUniversity Press, 1959.

[56] Margaret A. Peteraf, Jay B. Barney, Unraveling the resource-basedtangle, Manag. Decis. Econom. 24 (4) (2003) 309–323.

[57] Coimbatore K. Prahalad, Gary Hamel, 1990, The core competence ofthe corporation, in: Resources, Firms, and Strategies: A Reader in theResource-Based Perspective, Oxford University Press, pp. 235–256.

[58] Gautam Ray, Jay B. Barney, Waleed A. Muhanna, Capabilities,business processes, and competitive advantage: choosing thedependent variable in empirical tests of the resource-based viewhttp://doi.wiley.com/10.1002/smj.366, Strateg. Manag. J. 25 (1)(2004) 23–37, http://dx.doi.org/10.1002/smj.366.

[59] Indulska Recker, Jan Marta, Michael Rosemann, Peter Green, Theontological deficiencies of process modeling in practice, Eur. J. Inf.Syst. 19 (5) (2010) 501–525, http://dx.doi.org/10.1057/ejis.2010.38.

[60] Jan Recker, Michael Rosemann, Peter Green, Marta Indulska, Doontological deficiencies in modeling grammars matter? MIS Quar-terly. 35 (1) (2011) 57–79.

[61] Michael Rosemann, Peter Green, Marta Indulska, 2004, A referencemethodology for conducting ontological analyses, in: ConceptualModeling—ER 2004, Springer, pp. 110–121.

[62] Mari Sako, Supplier development at Honda, Nissan and Toyota:comparative case studies of organizational capability enhancement,Ind. Corp. Change 13 (2) (2004) 281–308, http://dx.doi.org/10.1093/icc/dth012.

[63] Paulo Sérgio Santos Jr., João Paulo A. Almeida, Giancarlo Guizzardi,An ontology-based analysis and semantics for organizational struc-ture modeling in the ARIS method, Inf. Syst. 38 (5) (2013) 690–708,http://dx.doi.org/10.1016/j.is.2012.09.004 ⟨http://linkinghub.elsevier.com/retrieve/pii/S0306437912001263⟩.

[64] Paulo Se ́rgio Santos Jr., João Paulo A. Almeida, Giancarlo Guizzardi,2010, An ontology-based semantic foundation for ARIS EPCs, in:Proceedings of the 2010 ACM Symposium on Applied Computing,pp. 124–130. ⟨http://dl.acm.org/citation.cfm?id=1774114⟩.

[65] August-Wilhelm Scheer, 2000, ARIS - Business Process Modeling.ARIS Business Process Modeling. Vol. 2., comple. ⟨http://books.google.com/books?id=Jzuwm2uV7CQC&pgis=1⟩.

[66] John R. Searle, Intentionality: An Essay in the Philosophy of Mind,Cambridge University Press, 1983.

[67] Graeme Shanks, Elizabeth Tansley, Jasmina Nuredini, Daniel Tobin,Ron Weber, Representing part–whole relations in conceptual model-ing: an empirical evaluation, MIS Q. 32 (2008) 553–573. doi:Article.http://csz.csu.edu.tw/pp/98/MISQ/2008/2008-3/REPRESENTING PART-WHOLE RELATIONS IN CONCEPTUAL MODELING.pdf.

[68] Janis Stirna, JanisGrabis, Martin Henkel, Jelena Zdravkovic. 2012.Capability driven development—an approach to support evolvingorganizations, in: The Practice of Enterprise Modeling, Springer,pp. 117–131.

[69] David Teece, Gary Pisano, The dynamic capabilities of firms: anintroduction, Ind. Corp. Change 3 (1994) 537–556, http://dx.doi.org/10.1093/icc/3.3.537-a.

[70] The Open Group, 2009, TOGAF Version 9, Architecture. ⟨http://books.google.com/books?hl=en&lr=&id=dNIf8TZaylcC&oi=fnd&pg=PT2&dq=TOGAFþVersionþ9&ots=W6dnwd5wtj&sig=C4emZBzdp5pkoMOeE9sy1FtaYX4⟩.

[71] Yair Wand, Ron Weber, 1989, An ontological evaluation of systemsanalysis and design methods, in: Information System Concepts: An In-Depth Analysis, Elsevier Science Publishers BV, Elsevier North-Holland.

[72] Yair Wand, Ron Weber, 1990, Mario Bunge's ontology as a formalfoundation for information systems concepts, in: Studies on MarioBunge's Treatise, Rodopi, Atlanta, Rodopi pp. 123–149.

[73] Ron Weber, Ontological Foundations of Information Systems, Coop-ers & Lybrand and the Accounting Association of Australia and NewZealand, Melbourne, 1997.

C.L.B. Azevedo et al. / Information Systems 54 (2015) 235–262262

[74] Robert R. Wiggins, Timothy W. Ruefli, Schumpeter's ghost: ishypercompetition making the best of times shorter? http://doi.wiley.com/10.1002/smj.492, Strateg. Manag. J. 26 (10) (2005)887–911, http://dx.doi.org/10.1002/smj.492.

[75] Faramak Zandi, Madjid Tavana, A multi-attribute group decisionsupport system for information technology project selection, Int. J.Bus. Inf. Syst. 6 (2) (2010) 179–199.

[76] Jelena Zdravkovic, Janis Stirna, Martin Henkel, JanisGrabis, 2013,Modeling business capabilities and context dependent delivery bycloud services, in: Advanced Information Systems Engineering,Springer, pp. 369–383.