Building a responsibility model including accountability, capability and commitment

Building a Responsibility Model including
Accountability, Capability and Commitment
Christophe Feltus1 and Michaël Petit2
1
Public Research Centre Henri Tudor, 29, Avenue John F.Kennedy, L-1855 Luxembourg-Kirchberg, Luxembourg
christophe.feltus@tudor.lu
2
Computer Science Department, University of Namur, B-5000Namur, Belgium
mpe@info.fundp.ac.be

This paper aims at building a responsibility model based on the concepts of Accountability, Capability and Commitment. The
model’s objectives are firstly to help organizations for verifying the organizational structure and detecting policy problems and
inconsistency. Secondly, the paper brings up a conceptual framework to support organization for defining their corporate, security
and access control policies. Our work provides a preliminary review of the researches performed in that field and proposes, based on
the analyses, an UML responsibility model and a definition of all its concepts. Thereafter, to propose a formal representation of the
model, we have selected the suitable language and logic system. The analyze highlights that an important variable is whether the
responsibility is perceived at a user or at a company level.

Index Terms—Responsibility, Capability, Commitment, Accountability, Access control, Right management, Formal system,
Security management.

by using that framework, verifying organizational structure
I. INTRODUCTION and detecting policy problems and inconsistency.

I t is notable that nowadays, the responsibility committed Our work will be based on the hypothesis that this
from a person to perform a task is an aspect that has for a responsibility is composed by the tuple (Capability,
long time remained overshadowed and that nevertheless Accountability, Commitment). Our previous work [2] has
appears to be from a major interest. The perception of introduced principal semantic characteristics about those three
responsibility has often been limited to a combination of rights concepts and has brought formalizing elements using standard
and obligations. However current business (for example in the logics.
financial sector) demonstrates that the moral aspect is The work is introduced by Camerer’s observations over
improvable and that taking care of that matter would avoid in research in the field of policy. These observations presented in
some cases malfunctions of the system. In practice, the next section provide a precious warning we have to take
responsibility is most often translated through policies. It care for our research. Section 3 reviews the concepts of
exists much definition of policy. For our work, we prefer the responsibility in access control models and in engineering
definition of policies from [31] that is Policies are rules that methods. Section 4 formalizes the responsibility and its
governe the choice in behaviour of a system. Security policies concepts with an UML model and presents the selection of a
define what actions are permitted or not permitted for what or formal system. Section 5 introduces future works around the
for whom, and under what condition (…) This definition is formalization and section 6 concludes.
interesting it that, even if it is coming from a low level The results presented in this paper are a contribution from
context, it sounds applicable to the high level one such as the the SIM (Secure Identity Management) project [36] and RED
management. (Reaction After Detection) project [43].
Based upon the above observations, the first objective of
that paper is to perform a literature review of policy models II. FROM BUSINESS TO SECURITY POLICY
and engineering methods to identify the main policy’s Before going ahead in the literature review, let make a
concepts. From that literature review, a model of hook to understand the analysis made by Camerer [9] on
responsibility is elaborated and integrates main researches in business policy and strategy. An important
responsibility’s concepts and main relationships between observation in his work is that: « There are at least three
those concepts. The specificity of that model is its genericity symptoms of the disease causing the queasy dissatisfaction
that permits in the first hand to integrate policies from all with policy research:
abstraction layers of the company, e.g.: IT policy are declined 1. Concepts are often ambiguous and their
from Corporate policy, and in the second hand, to be definitions are not agreed upon;
compatible to policy from different domains of the company.
E.g.: IT policy, organizational policy, or security policy. 2. Checklists or theories are rarely tested, and never
Finally, we introduce a formalization of the concepts using tested directly against competing theories and
logic system. The formalization main objectives are to 3. Theories do not ‘cumulate’ or built upon previous
propose a basic logic framework for defining all concepts and, theories as they should.
These three deficiencies are a result of the way policy

research is typically done.” been produced in that domain [15][17][18] and [19] but none
Camerer explains that policy research should evolve from has targeted the responsibility through the tuple (Capability,
an inductive to a deductive approach. He argues that induction Accountability, Commitment).
contribute to an unproductive debate about variable Despite that proliferation of works, it is noteworthy that
definitions and to a lack of testability and failure of theory. up to now there does not really exist a distinction between
Unlikely, his conviction is that deductive models can express works addressing access control model, policy model, role
hypotheses in a language that is more amenable to progressive engineering and permission/policy engineering. Based on that
debate. This point of view is a precious warning we have to assumption, it appears meaningful for apprehending that topic
take into account before beginning our researcher in that it to clarify this point and to highlight the existing dichotomy
may prevent us to perpetrate the same mistakes. This warning between model and method. To perform our review, we will
is moreover substantial because of the subjective character of base our analysis on a commonly accepted idea that a model
the moral aspect under focus in our research. In his work, or conceptual model is a representation designed to show the
Camerer only addresses business policy. Therefore, this structure of a system or concept and that (at least in our case),
consideration needs to be adapted according to our research’s a method is a body of techniques for collecting data necessary
context and it is consequently necessary to clarify the relation to instantiate the conceptual model. Consequently and as
that exists between business policies and IT policies. Wies illustration, the Role-Based Access Control (RBAC) model
[10] shows the links between high and low-level policies. He [1] proposes a structure for providing access based on role
depicts the variation of importance of the technology and the whereas role engineering [3] and [4] is a method aiming to
business aspects when translating high-level onto low-level define roles to instantiate the conceptual model. Identically,
policies. High-level policies tend to focus on business aspects policy may also be modeled and there exists a proliferation of
whereas low-level policies focus on technology aspects. methods to instantiate it. These methods may be classified
Although they are spread on different abstraction layers of the according to the technique they use. We propose to start with
policy hierarchy, business policies and IT policies should be methods based on Requirements Engineering (RE) and to
consistent because both should be derived from (management continue with a list of others. Moreover, it is more frequent to
and/or IT) goals and hence embody (management and/or IT) read paper targeting policy language than policy model. Those
strategy’s aspects. Rifaut et al. [11] propose to use Goal- policy languages are innumerable and spread over the entire
Oriented Requirements Engineering (GORE) methods to organizational model layers. Most famous of them are Ponder
define goals, strategies and policies. Rifaut explains that these [5], Policy Description Language [6], Security Policy
methods can be used to analyze and model systems at all Language [7], and Rei [8]. Amazingly, the policy model used
organizational level, from business models down to to support the policy expression by the policy language
architectures, see Fig. 1. He argues that the four artifacts that remains rarely specified. This review presents successively the
are objectives, policies, strategies and indicators may be responsibility through access control models and engineering
globally considered as objectives and that consequently, low methods. The components of the responsibility’s tuple are :
level objectives contributes to achieve higher level one. E.g.: • Capability: which describes the quality of having the
Having access control management contributes to have a requisite qualities or accesses to resources to achieve a
performance IT security and having a performance IT security task;
contribute to have a good corporate governance. • Accountability: which describes the state of being
answerable about the achievement of a task;
• Commitment: which is the engagement of a stakeholder
St r
ate ine
lev gic B us lue
ss Policies
Goals

Strategies
Indicator s

to fulfil a task and the assurance he will do it.
e va
Ta c l
These definitions are refined through the description of
Goals

ss
Indicators

in e
Policies Str ategies

t
l ic a s
Bu ce sses
Op evel l
Goals

p ro these concepts in section IV. A.
Indicators

e ra s
Policies Strategies

t
l ev ion al ur e
ced :
ain …
Te c el P ro d o m ents,
l ev
hni
el
c al
.
he IT mpo n
in t s, co Objectives Responsibility in the field of IT has already been
E. g tion
app
l ica Concepts
Indicators investigated because of IT security constraints and
Standard view of Policies Strategies
organizational layers
and artifacts of
organizational layers requirements firstly, and of software requirement engineering
secondly. IT security depicts responsibility mainly when it
Fig. 1. GORE model for Policy refinement addresses access control. Indeed, to provision employees with
right and obligation to operate over an application or a
Based on the previous assumption that there exist links component, main access control model use the concept of role
between policies from different layers, further analysis of the to group employee based on their responsibility, function,
literature has been conducted to depict the principal elements geographic location, domain of work, etc. Some examples of
that compose the policy concept. those models are the Mandatory Access Control, RBAC [10],
UCON [11], OrBAC [12], etc. However, the inconvenient
III. RESPONSIBILITY LITERATURE REVIEW already observed in large company is that the engineering of
that roles leads sometime to situations where the amount of
It is rapidly observable when analyzing policy literature roles is bigger than the amount of employees.
that a very large amount of authors show interest in that
Responsibility has also been subject of research in the field
concern. Consequently, a number of surveys have already
of software requirement engineering. Indeed, this concept is

centric for a large amount of methods like I*[13]. I* makes Our survey has also covered others approaches that due to
goal-oriented strategic modeling and analysis of requirements the size of the paper are not presented here. In summary we
by using three mains concepts that are: actors, intentional may observe that firstly, some concepts are commonly
elements, and links. Actors are described in their accepted, such as right, role and obligation. Definition of the
organizational setting and have attributes such as goals, two firsts concepts are scarce. Only one definition has been
abilities, beliefs, and commitments. Actors can be agents, found for the concept of “right”: the right (or permission) is
roles, and positions. Agents are concrete actors, systems or explicitly granted to a subject to access an object in a specific
humans, with specific capabilities. The inconvenient of those mode, such as read or write [1]. For the concept of “role”,
methods is that they are limited to concepts directly linked to only one definition has also been found in [13]. The concept
of obligation is subject to more debate. For Bettini et al. [14],
the software requirement like the right or the obligation
obligations are conditions or actions that must be fulfilled
without offering the possibility to be extended to wider
either by the users or the system after a decision. In [1],
concepts like the commitment.
Sandhu et al. define obligations as requirements that have to
The state of the art of policy concepts introduces a review be fulfilled by the subject for allowing access. Crook et al.
of 4 main recognized access control models: Mandatory [15] extend the notion of obligation to obligation policy that
Access Control (MAC), Discretionary Access Control (DAC), relate to actions that must be carried out on targets by subjects
Role-based Access Control (RBAC) and Usage Control when a predefined event occurs and Haley et al. in [16] define
Model (UCON). it as what actions must be taken before access can be granted.

TABLE 1.
AC MODEL AND RESPONSIBILITY’S CONCEPTS

MAC DAC RBAC UCON
Subject Subject Subject User Subject
Object Object Object Object Object
Defined by objects and subject’s
Group No User Group Role
attributes
Capability Access Right Access Right Access Right Access Right
Accountability Yes, static and dynamic Defined by objects and subject’s
No No
(Obligation, Constraint) separation of duty attributes
Commitment No No No No

TABLE 2.
ENGINEERING METHODS AND RESPONSIBILITY’S CONCEPTS.

Scenario
KAOS I* GBRAM ARMF RACAF Uses Cases
Driven
Subject Agent Actors Agent Users Actors Subject Actors
Object Entité / model objet Yes - Asset Data - Object
Group - Yes - Role Yes Yes Yes
Abilities
Capability (Right,
Authorization rules and - Permission Permission Permission Access right
Authorzation)
beliefs
Accountability Achieve requirements Achieve Perform Perform Perform Pre-conditions,
Goal
(Obligation, Constraint) and expectations a goal a task a task a scenario post-conditions
Commitment No Yes No No No No No

Table 2 is a summary and a comparison of the reviewed IV. FORMALIZATION OF THE RESPONSIBILITY
engineering methods. We may observe that, because the most This chapter aims at defining a responsibility model to
frequently addressed concern of capability is the access right, clarify and better understand concepts that compose
existing models and methods most of the time remain responsibility notion. To achieve that, we firstly use the
targeting low-level layers of abstraction of the organization. Unified Modeling Language (UML) to represent the
Moreover, if we consider responsibility as a tuple (Capability, components of the model and their relations and then, we
Accountability, Commitment), we observe that nowadays propose to introduce the formalization of its components with
there exists no model and method that entirely take into formal language and logic system. With the desire to keep this
account all these responsibility components. Other paper didactic and to grant a common understanding of
responsibility models exist but are often links to social or responsibility concepts, the work will be grounded based on
psychological areas, or in very specific domains like [41, 42]. the following case study:

Mister Boss is the manager of the marketing company

named “SelltheWorld”. Each year, Mister Boss organizes facilities, those users are often grouped together based on
during the Christmas period a large sending of postcards to their profile. As previously explained in the literature
all its customers. This year, Mr Boss has too much work for overview, the most famous type of classification is the role
closing the annual report and consequently decides to but variations exist such as for example the team, the
delegate this task to one of its employees. Because the task is hierarchy, or some geographical constraints.
less business sensitive as some other production task, Mr
Boss decides to delegate it to a part-time secretary named
Sophie. Sophie has just get married and consequently, she
accepts this additional work without commitment. Mr Boss
asks to the IT service manager to give Sophie the necessary
access right to the customers address list. The IT service
manager asks an employee from the IT service named John to
realize the necessary operation for providing this right. On
January the 30th, Mister Boss receives over 100 complains of
customers that didn’t receive Christmas card.
Mr Boss has duly formalized Sophie’s Accountability by
asking her to realize the sending activity. It was consequently
clear about what she was accountable to do. To achieve that
sending, she got the necessary capability that was the access
to the customers file. However, due to the fact that her
thought went to her new husband rather that to the work to
accomplish, she didn’t really want to achieve the work and
failed to assure her responsibility due to a miss of Fig. 2. UML model of responsibility applied to the right management
commitment.
John’s responsibility can also be analyzed by that case
study. John is a well paid IT staff that is very happy with his • Role: Role describes the position of a person in the
function. He has received clear accountability to give access organisation. This position may be related to a hierarchical
right to Sophie and he has the needed capabilities due to its status, a geographic position, the membership to an
position as network administrator. He has consequently been organisation unit or department, or whatever. This component
responsible to fulfill Mr Boss’ request. is largely present in the literature that provides some
definitions of it.
A. Responsibility model • Responsibility: It also exists a plethora of definitions
This section presents our model of responsibility. The of responsibility and this paper has not for duty to propose a
major interest of it is its genericity. Indeed, the model aims to new one. We may however state that commonly accepted
be generic enough to be applied to all kind of organizations, responsible definition encompasses the idea of having the
at each abstraction layers of it, and also for all domains of the obligation to ensure that something happens. Moreover, the
company like for example the IT security (and the above literature review shows that it makes sense to hang on
management of access right), the management, or the to it the three additional elements that are Capability,
production. Accountability and Commitment. One basic relation existing
Some components of the model are generic in that they are in the model is consequently the relationship between
present at all instantiations of it. Others components have Responsibility and Capacity, Accountability and
been added with the objective to illustrate the application of Commitment. This relation is of the form 0..* to 1. That
the model in the context of the management of access rights. means that being responsible involves that it is possible to
Those components are “Access Right” and “Resource”. Our dispose of many Capacities, Accountabilities and
model reuses some commonly accepted components Commitment. But at the opposite, on Commitments is only
presented in the literature survey in sections 3 and 4, whereas bound to one responsibility, and adequately for
others are new. The model encompasses the following Accountability and Capability.
concept: • Task: is the operation performed by the role (or the
• Organization: At the top of the UML model (see Fig. user), which is responsible for it. This concept doesn’t exist in
2) is the organization. Organization represents a structure that the realm of access control model that tends rather to speak
pursues collective goals and that is limited by a defined about right or/and obligation needed to perform an operation.
border. This structure encompasses employees (users) that are E.g.: The right to read a document or the obligation to satisfy
responsible to perform tasks (or processes) that implicitly conditions before executing an operation. By contrast, task is
generate profit. Organization also encompasses resources that a centric concept in requirement engineering. E.g.: in Tropos,
could be whether produced by the task or used by a user to a goal may by achieve by fulfilling a task. The relation
perform a task. between role, responsibility and task is to be underlined. This
• User: User appears as a person external or internal to relation is to be read: “there is one and only one role
an organization, a system or a software component. User has responsible for one task, and one role may have many
to achieve a task he is responsible for. Number of synonyms responsibilities and one responsible may perform many
of it exists like subject, actor or agent. For administration tasks”.

• Accountability: is a concept that exists mainly in activity to achieve because he does not shared the result
engineering methods and that appears through the obligation anymore.
to achieve a task or to perform an action. This concept • It addresses the commitment aspect of the
describes the state of being answerable about the achievement responsibility and consequently increases the ethics of the
of a task. The case study above illustrates that Sophie is business in general.
accountable toward Mr Boss regarding the task she has been • It guarantees that the right capability is affected to the
assigned responsible for. In the same way, John is right user. This advantage guarantees that the responsible
accountable toward the IT manager for providing the access receive the minimum privilege necessary for achieving the
right. task and consequently, it limits the vulnerability of the
• Commitment: is the moral engagement of a system.
stakeholder to fulfil a task and the assurance that he will do it.
B. Selection of a formal system
Commitment is the most infrequent concept. Traditional
policy model such as RBAC do not address it, however i* Even if this model brings up a first contribution for
partly introduces it (e.g. when defining dependency as an verifying the organization structure and detecting policy
“agreement” between two actors). However, to distinguish if problems and inconsistency, it appears impossible to exploit it
it is a moral concept or an obligation remains interpretable. without the help of a formal language. This section introduces
This component is illustrated through the cases study as a preliminary reflection over the selection of that language.
follow: Firstly, we may state that because Sophie has other To select a language, we may state that the model of
duty in mind, she has not the willingness to achieve the task. responsibility formalizes information that represents
We may state that she is not committed to do it. At the responsibility elements in force in the company. That
opposite, John is a well paid IT staff that is very happy with information composes a system that is part of the real world
his function. He is fully committed to perform the task. called the universe of discourse and that encompasses a
• Capability: which describes the quality of having the number of properties (constraints) that the system must
requisite qualities, skills or resources to perform a task. satisfy. In [38], Meyer at al. explains that some of the
Capability is a component that is part of all models and constraints may not be violated and could be formalized using
methods, and is most frequently declined through definition predicate logic, temporal logic or dynamic logic whereas
of access rights, authorizations or permissions. Based upon others are violable and formalized using deontic logic. The
the above case study, the Capability is illustrate through the constraint that before to have access to a file, it is necessary
Sophie’s capability to access the customer’s file. This that the right for accessing the file has been dully set on the
Capability exists because John was responsible to provide that fileserver is inviolable. Indeed, according to our case study, it
access right. The case study illustrates also John’s Capability is impossible that Sophie get access to the customers list if
to be responsible for providing access right. Indeed, due to his she doesn’t have the right to read the concerned file. If we
position of network administrator, he has the right to manage consider that read the file is a proposition, we can deduce that
all employees’ access right. having the access right is a capability or a modal operator of
Additionally, the UML model of responsibility (Fig. 2) “read the file”. Some others constraints are considered as
includes two added elements to the basic responsibility ideal but violable. This could be illustrated by the
model: “access right” and “resources”. These elements permit responsibility of John (that as to set the necessary access right
to illustrate the case of a particular type of Capability that is for Sophie) that due to an overload of work did not have
the access to resources. We define a resource as something enough time to achieve it. Time is considered in that example
needed for or produced by performing a task and that can as the capability necessary to fulfill the task. John has not
takes a large scale of representation such like information, assumed is responsibility because the statement that John is
manpower or money. The access right is defined as a capable to do it has been violated.
statement over the type of action that could be performed by a In [33], Cholvy et al. propose a formalization of the
user over that resource. This access right is a Capability for a concept of responsibility. In her work, she explains that
responsible while being at the same time accountability for responsibility is a concept that has several facets that
another. This relationship between Capability, access right correspond to very different meanings. She extracts three
and accountability has been more deeply explained in [2] and definitions of responsibility, which implicitly encompasses
[36]. In our model, Capability is a broader concept than the the three concepts from our model (capability, accountability,
mere one of access right. commitment). The first definition links the responsibility
concept to something bad that has happened to a person that
The advantages of such a model (Fig. 2) are important for
could have caused or prevented it. This definition is mainly
4 reasons:
issued from the legal world. The second definition issued
• It permits to improve the business/IT alignment and from Cholvy’s paper claims that responsibility is an
brings material to answer to the principle 1 of the ISO/IEC obligation or a moral duty to report or explain the action or
38500:2008 standard [40]: Establish clearly understood someone else’s action to a given authority (answerability).
responsibilities for IT. This definition helps at defining the commitment as a moral
• The accountability is bound to the agent rather than to duty in parallel with an obligation that is considered as a legal
a group of agents (like in others models [39]). This makes the duty. The third definition defines the responsibility according
agent personally more involved and more concerned by the to a position in an organization and explains that someone

responsible for something should be prepared to justify his
action. This justification brings the content of the concept of
accountability and consequently nuances accountability
versus answerability. Based upon the three definitions,
Cholvy proposes a logic framework and explains how the
framework may be used to model different aspects of the
responsibility. She used the deontic logic and the logic of
actions to achieve that. Deontic logic is the field of logic that
is concerned with obligation (O), permission (P) and
prohibition (F), and that permits to reason about ideal versus
actual states or behavior.
According to her approach, and based upon the Meyer’s
explanations over the necessity to prefer deontic logic for
modeling system that encompasses ideal but violable
properties, we may rightly agree that Cholvy’s choose is
suitably justified. If we consider the model of responsibility
as a user based representation of one responsibility, what
means in other words, that the concepts of responsibility
introduced in the model represents the responsibility of a
unique user to perform a unique task, the three components
that compose the responsibility tuple are violable. For
example, regarding the capability, we may state that based
upon the case study, Sophie must have the list of addresses.
However, it may happen that due to undefined reasons, she Fig. 3. UML model of multiple responsibilities interactions
doesn’t have it.
If we expand the sphere of responsibility from a user While based on our hypotheses that the existence of
based perception to an organization based perception, this capability and accountability is inviolable, the concept of
statement is no more automatically true. Indeed, if the commitment is more likely to discussion. Because this
company is considered as a set of tasks, persons, and concept is strongly depending of the moral willingness, we
responsibilities, we may suppose that in an ideal situation, it may argue that no real elements may absolutely guarantee its
must exist at least one Capability, one Accountability and one inviolability. This affirmation may however be nuanced if we
Commitment corresponding to each responsibility. look toward social, psychology, or managerial sciences. The
The existence of capability and accountability concepts is salary, the relationship with colleagues, or the concordance of
easily manageable and verifiable. Indeed, it is easy for an the job with the interest of the employee are some elements
operation or a processes manager to determine the dully that probably influence it. However, in this paper we consider
capabilities necessary to perform a task or to clearly fix the that those elements are not objectively manageable and do not
expected accountabilities. Moreover, such concepts are easily provide a guarantee of inviolability. We will consequently
traceable in a database for example or with a software tool. prefer the usage of deontic logic for formalizing that element.
This exercise has already been achieved in previous works We may consequently suppose that some elements of
[36]. The consistency between both concepts may also be responsibility may be formalized using predicate logic and
examined based upon the supposition that the capability others with deontic logic.
needed for assuming a responsibility corresponds to
accountability of another user’s responsibility. V. FUTURE WORKS REGARDING THE FORMALIZATION OF THE
Fig.3 illustrates that links. Based upon our cases study, we RESPONSIBILITY
may consider that : Additionally to the Cholvy’s proposition to formalize
a) Sophie’s capability (having access right) is the responsibility with deontic logic and action logic, our future
accountability of John (provides access right). works extend the formalization of the responsibility with the
b) John’s capability (having time for performing the components of the responsibility tuple (Capability,
right management) is the accountability of the IT Accountability and Commitment). The responsibility (R)
service manager (provide time the IT service staff) assigned to a user (u) to perform a task (t) is written R([t]u).
c) IT service manager’s capability (having budget to Based upon our previous observations, we state that this
hire IT employees) is the accountability of Mr Boss formalization has one specificity that resides in that the
(provide IT service manager budget) components of the responsibility’s tuple are at the same time
If we base our work on that reasoning that in an ideal conceptual components and modal operators: capability (CA),
situation, responsibilities in a company are dully fixed and accountability (AC) and commitment (CO). We have
that capabilities and accountability exist for each consequently to develop a formalization based on the deontic
responsibility, we may conclude that those two concepts are logic to formalize the user based formalization of the
inviolable and may be formalized using predicate logic. responsibility and extend this formalization to predicate logic

to represent the responsibility at an organization level. An Whatever, not achieving a task for which the user is
envisaged possibility to define responsibility’s modal accountable may lead to some kind of blame. This aspect is
operators is to develop the user based representation of the not discussed in that paper.
responsibility based on the adaptation of the Traditional Future formalization works will also aims at defining the
Threefold Classification (TTC) [37]. To achieve that, we commitment. We already suppose that it will be necessary to
transpose Obligatory to Accountable in that both modal also define it based on the TTC. Fig. 5 shows how it seems
operators bring up the notion of a constraint that is logic to represent it.
indispensable and makes obligatory by a legal issue (e.g.: a
policy), we transpose Permissible by Capable in that both VI. CONCLUSIONS
defend the idea that this constraint permits an action to be We have analyzed the literature to understand the
performed. And we keep the Optional (OP) modal operator of semantics of AC policy conceptual models and engineering
the standard deontic logic unchanged. To achieve that methods. We have observed that some elements are
transposition, we need to define the Incapacity (IN) and the commonly accepted components whereas others remain
Unaccountability (UN) (see 2’ And 3’). Moreover, equally to debated or not addressed. Commonly accepted concepts are
the deontic standard schema, the Fig. 4 highlights that the user (and related ones such as group or role), resource and
three rectangular cells are jointly exhaustive and mutually capability. Capability is most frequently declined under
exclusive. Indeed, each proposition is accountable, optional or access right, authorizations or permissions. Accountability is
incapable. Moreover, Capable modal operators are those that a concept that exists mainly in engineering methods and that
are either Accountable or Optional and Unaccountable modal is declined as the obligation to achieve a task or to perform an
operators are those that are either Optional or Incapable. action. Commitment is the most infrequent concept. Based
upon that observation, we have developed a conceptual model
of responsibility using an UML class diagram and have
defined all the conceptual components and clarified some
important relationships between those. Thereafter, to propose
a formal representation of the model, we have selected the
suitable language and logic system. The analyze highlights
Fig. 4. From Traditional toward a Responsibility based Threefold
Classification that an important variable is whether the responsibility is
perceived at a user or at a company level.
Based upon the TTC, the Traditional Definitional Scheme In this paper, the responsibility concept has mainly been
(TDS) [37] states by the set of definitions from 1 to 4 that addressed based on an IT approach. However, the
something is permissible if and only if its negation is not “operational” and “management” fields are also rich of
obligatory, impermissible if and only if its negation is responsibility’s theories [34] and [35]. This area will be the
obligatory, gratuitous if and only if it is not obligatory, and focus of our future researches and will permit to refine our
optional if and only if neither it nor its negation is obligatory. first findings. Consequently, our future works will focus on
If we consider that the proposition (p) is the performance of a continuing the development of the model of responsibility,
task (t) by a user (u) and is noted [t]u, the set of definitions and most specially the concept of commitment that is
from 1’ to 4’ may defines the concepts of the responsibility important to consider in high-level layer of the organizational
according to the Responsibility based Threefold model. Moreover, defining policy that allows taking into
Classification. account the commitment opens doors to new approaches that
have right now poorly be taken into account in traditional and
PEp ↔ ~OB~p (1) renowned risk management solutions
IMp ↔ OB~p (2) As a conclusion regarding the Camerer’s warning of
GRp ↔ ~OBp (3) section I we have done this analysis to clarify the semantic of
OPp ↔ (~OBp & ~OB~p) (4) all components that encompass the responsibility and we may
consequently state that symptom 1 and 3 identified by
CA[t]u ↔ ~AC~ [t]u (1’) Camerer has been addressed. Firstly the symptom 1 that is
IN[t]u ↔ AC~ [t]u (2’) “Concepts are often ambiguous and their definitions are not
UN[t]u ↔ ~AC[t]u (3’) agreed upon” has been partially tackled with clear literature-
OP[t]u ↔ (~AC[t]u & ~AC~ [t]u) (4’) based enlightenment of the concepts. Secondly symptom 3
that is “Theories do not ‘cumulate’ or built upon previous
For achieving a task, u must have the necessary theories as they should.” has been addresses with a tentative
capabilities and be committed to perform it. Whether or not definition of “responsibility” considering the way its
he is accountable do not presents any impact on the conceptual component are addresses by others authors.
realization. Another part of our work aims at defining a new approach
to derive the responsibility from the high-level down to the
lower one. Our first researches demonstrate that potentials
solutions are to link responsibility’s concepts with
organization’s processes. To support the progress of that
approach, a software prototype has been developed based on
“egroupware open framework”. Those researches and the
Fig. 5. Commitment on Responsibility Threefold Classification prototype have been presented in [36].

REFERENCES [24] Robert Crook, Darrel Ince, Bashar Nuseibeh, Towards an Analytical
Role Modelling Framework for Security Requirements, Security
[1] R. Sandhu, J. Park, Usage Control: A Vision for Next Generation Requirements Group, Departement of Computing, The Open University,
Access Control, The Second International Workshop on Mathematical Walton Hall, Milton Keynes, MK7 6AA, UK.
Methods, Models and Architectures for Computer Networks Security, [25] Henry Mintzberg, Structure in Fives: Designing Effective
2003. Organisations, Englewood Cliffs, NJ: Prentice-Hall, 1983. pp. 312
[2] C. Feltus, A. Rifaut, An Ontology for Requirements Analysis of [26] Qingfeng He, Annies I. Antón, “A Framework for Privacy-Enhanced
Managers’ Policies in Financial Institutions, I-ESA2007, Madeira, Access Control Analysis in Requirements Engineering”, Proc. of the 9th
Portugal. International Workshop on Requirements Engineering: Foundation for
[3] Gustaf Neumann, Mark Strembeck, A Scenario-driven Role Engineering Software Quality (REFSQ'03), pp. 137-146, Klagenfurt/Velden, Austria,
Process for Functional RBAC Roles, SACMAT’02, June 34, 2002, June 16-17, 2003.
Monterey, California, USA. [27] E. B. Fernandez and J. C. Hawkins, “Determining Role Rights from Use
[4] Coyne, E. J. 1996. Role engineering. First ACM Workshop on Role- Cases”, Proc. of the ACM Workshop on Role-Based Access Control,
Based Access Control, Gaithersburg, Maryland, United States. 1997.
[5] N. Damianou, N. Dulay, E. Lupu, M. Sloman , The Ponder Policy [28] Roeckle, H., Schimpf, G., and Weidinger, R. 2000. Process-oriented
Specification Language Workshop on Policies for Distributed Systems approach for role-finding to implement role-based security
and Networks (Policy2001), HP Labs Bristol, 29-31. Springer-Verlag. administration in a large industrial organization. In Proceedings of the
[6] Bertino, E., Mileo, A., and Provetti, A. 2005. PDL with Preferences. Fifth ACM Workshop on Role-Based Access Control (Berlin, Germany,
IEEE international Workshop on Policies For Distributed Systems and July 26 - 28, 2000). Role-Based Access Control '00.
Networks, Policy 2005 – Vol. 00, IEEE Computer Society, Washington, [29] Chandramouli, R. 2001. A Framework for Multiple Authorization Types
DC, 213-222. in a Healthcare Application System. 17th Annual Computer Security
[7] Basile, C.; Lioy, A.; Perez, G. Martinez; C., F. J. Garcia; Skarmeta, A. Applications Conference, 2001. ACSAC. IEEE Computer Society,
F. Gomez, POSITIF: A Policy-Based Security Management Washington, DC, 137.
SystemPolicies for Distributed Systems and Networks, 2007. [30] D. J. Thomsen, Richard C. O'Brien and C. Payne, Napoleon: Network
POLICY’07, pp. 280 – 280. Application Policy Environment, ACM Workshop on Role-Based
[8] Lalana Kagal, Rei : A Policy Language for the Me-Centric Project, Access Control, 1999, pp. 145-152.
TechReport, HP Labs, September 2002. [31] N. Dulay, E. Lupu, M. Solman, N. Damianou, A Policy Deployment
[9] Colin Camerer, Redirecting Research in Business Policy and Strategy, Model for the Ponder Language , An extended version of paper in Proc.
Strategic Management Journal, Vol.6, No. 1. (Jan. – Mar., 1985), pp. 1- IEEE/IFIP International Symposium on Integrated Network
15. Management, (IM’2001), Seattle, May 2001, IEEE Press.
[10] René Wies, Using a Classification of Management Policies for Policy [32] OASIS, “eXtensible Access Control Markup Language (XACML)
Specification and Policy Transformation. In Proc. ISINM '95, Santa Version 2.0” February 2005. www.oasis-open.org/committees/xacml/
Barbara, California, May 1995. [33] L. Cholvy, F. Cuppens, and C. Saurel. Towards a logical formalization
[11] André Rifaut, Christophe Feltus, Improving Operational Risk of responsibility. In Proc. of the Sixth International Conference on
Management Systems by Formalizing the Basel II Regulation with Goal Artificial Intelligence and Law, pages 233--242, 1997.
Models and the ISO/IEC 15504 Approach, REMO2V’2006, [34] Mintzberg H. Mintzberg on Management: Inside our strange world of
Luxembourg. organizations. The Free Press, New York, 1989.
[12] Davrondhon Gafurov, Kirsi Helkala, Nils Kalstad Svendsen, Security [35] Gray, B. Collaborating. Jossey-Baas, San Francisco, 1991.
models for electronic medical record, Telektronikk 1.2005. [36] J. Aubert, B. Gateau, C. Incoul, C. Feltus, SIM : An Innovative
[13] David F. Ferraiolo, Ravi Sandhu, Serban Gavrila, D. Richard Kuhn and Business-Oriented Approach for a Distributed Access Management,
Ramaswamy Chandramouli, Proposed NIST Standard for Role-Based International Conference on Information & Communication
Access Control, ACM Transactions on Information and System Technologies: from Theory to Applications (IEEE ICTTA2008),
Security, Vol. 4, No. 3, August 2001, Pages 224-274. Damascus, Syria.
[14] C. Bettini, S. Jajodia, X. S. Wang, and D. Wijesekera, Provisions and [37] http://plato.stanford.edu/entries/logic-deontic
Obligations in Policy Management and Security Applications, 28th [38] J.-J. Ch. Meyer, R.J. Wieringa, F.P.M. Dignum, The Role of Deontic
VLDB conference, China, 2002. Logic in the Specification of Information Systems, International Series
[15] Robert Crook, Darrel Ince, Bashar Nuseibeh, Modelling access policies In Engineering And Computer Science archive, Logics for databases
using roles in requirements engineering, Information and Software and information systems book contents, Kluwer, pp. 71-115, 1998.
Technology 45 (2003) 979-991. [39] C. Feltus, Preliminary Literature Review of Policy Engineering
[16] Charles B. Haley, Robin C. Laney, Jonathan D. Moffett, and Bashar Methods- Toward Responsibility Concept, ICTTA2008, Damascus,
Nuseibeh, Using Trust Assumptions with Security Requirements, Syria.
Requirements Engineering Journal, vol. 11 no. 2 (April 2006) pp. 138- [40] International Standard for Corporate Governance of IT (IT Governance)
15. - ISO/IEC 38500, 2008
[17] Robert Crook, Darrel Ince, Bashar Nuseibeh, On Modelling access [41] I. Sommerville, T. Storer, and R. Lock. Responsibility modelling for
policies: Relating Roles to their Organisational Context, RE 2005, Paris. contingency planning. In Workshop on Understanding Why Systems
[18] Pete A. Epstein, Engineering of Role/Permission Assignement, PhD Fail, Contingency Planning and Longer Term Perspectives on Learning
thesis. from Failure in Safety Critical Systems, June 2007.
[19] Crook, R., Ince, D., and Nuseibeh, B., “Using i* to Model Access [42] [42] P.M Wright, K. White, D. Gaebler-Spira (2004). Exploring the
Policies: Relating Roles to their Organisational Context”, Social relevance of the personal and social responsibility model in adapted
Modelling for Requirements Engineering, Giorgini, P., Maiden, N., physical activity: A collective case study. Journal of Teaching in
Mylopoulos, J., and Yu, E., eds., MIT Press, 2006. Physical Education, 23(1), 71-87.
[20] P.J. Fontaine, Goal-Oriented Elaboration of Security Requirements. [43] B. Gâteau, D. Khadraoui, C. Feltus, B. de Rémont, Multi-Agents based
M.S. Thesis, Dept. Computing Science, University of Louvain, June Architecture for IS Security Incident Reaction, 6th IEEE International
2001. Conference on Computer Science, 2008 IEEE International Conference
[21] Yu, E. S. and Liu, L. 2001. Modelling Trust for System Design Using on Research, Innovation & Vision for the Future (IEEE RIVF 2008),
the i* Strategic Actors Framework. Workshop on Deception, Fraud, and 14-16/7/2008, Ho Chi Minh City, Vietnam.
Trust in Agent Societies Held During the Autonomous, Eds. Lecture 35
194.
Manuscript received September 30, 2008. Corresponding author: C. Feltus
[22] L. Liu, E. Yu, J. Mylopoulos, Analyzing Security Requirements as
(e-mail: christophe.feltus@tudor.lu).
Relationships Among Strategic Actors, SREIS’02, Raleigh, North
Carolina, 2002.
[23] Antón, Goal-Based Requirements Analysi,. Second ICRE’96, Colorado
Springs, USA, 1996.

Building a responsibility model including accountability, capability and commitment

More Related Content

What's hot (8)

Viewers also liked (8)

Similar to Building a responsibility model including accountability, capability and commitment (20)

More from Luxembourg Institute of Science and Technology (20)

Recently uploaded (20)

Building a responsibility model including accountability, capability and commitment