Towards an Ontology of Task Dependence in Organizations
Mena Rizk
a
, Mark Fox
b
and Daniela Rosu
c
Enterprise Integration Laboratory, University of Toronto, 5 King’s College Road, Toronto, Canada
Keywords:
Coordination, Enterprise Modelling, Ontology, Organization Model, Task Dependence.
Abstract:
In the face of an increasingly dynamic, complex, and uncertain task environment, effective coordination is
crucial for organizational success. Based on a real-world case study investigating the nature of coordination
challenges in a municipal infrastructure project, we identified shortcomings in the representational frame-
works offered by organization studies and enterprise modelling that limit their ability to effectively model task
dependence and assist in improving coordination. Starting from existing organizational research literature
and domain expertise, we conducted an ontological analysis of task-related concepts, formulated represen-
tational requirements, and proposed a formalization. Our approach defines task dependence in terms of the
constraints one task imposes on another, underpinned by novel constructs that define how and why a task is
constrained. These constructs support the inference of dependencies between tasks, facilitating the discov-
ery of potentially hidden, latent dependencies. We formalized our conceptualization in an ontology, detailed
herein using first-order logic. Consistency-verified implementations in Prover9 and OWL are provided. We
validated our approach by modelling and solving real-life scenarios provided by our domain-expert collabora-
tors. Our approach lays the groundwork for future extensions that will tackle the modelling of different forms
of dependence between agents within an organization.
1 INTRODUCTION
This paper introduces a novel model of task depen-
dence designed to assist organizations in overcom-
ing coordination challenges in complex, dynamic, and
unpredictable work environments.
The focus of this paper is on task dependencies —
the dependencies between tasks performed by agents,
rather than the agents themselves. While acknowl-
edging the existence of other forms of dependence
like epistemic, reward, and outcome (Puranam et al.,
2012; Raveendran et al., 2020), we intentionally ini-
tiate our endeavour at the task level with the aim to
robustly represent the inherent circumstances that ne-
cessitate coordination, and the tangible consequences
of dependencies on task execution. This allows for the
modelling of the irreducible dependencies that pertain
to the nature of work undertaken by an organization,
and the subsequent implications on the integration of
work, irrespective of agents’ work assignment.
We grounded our efforts in a case study (Section
2) conducted with one of North America’s ten most
a
https://orcid.org/0009-0008-6095-0100
b
https://orcid.org/0000-0001-7444-6310
c
https://orcid.org/0000-0002-5877-9681
populous cities, identifying coordination challenges
within a large-scale infrastructure project, and ana-
lyzing the representational requirements for reason-
ing about their resolution. In Section 3.1, we present
an ontological analysis of task, task structure, and
task dependence concepts and introduce definitions
for each concept that are informed by both the ex-
isting organizational literature and the use cases pro-
vided by domain experts. A review of the relevant
literature (provided in Section 3.2) discusses the rep-
resentational limitations of the existing frameworks.
Subsequently, we present our working conceptualiza-
tion of task and task dependence, in Section 4. Its
formalization in first-order logic is shown in Section
5. We validated our approach’s capacity to represent
dependencies and demonstrated its reasoning affor-
dances by applying it to scenarios from a real-life case
study (Section 6). We discuss its advantages and lim-
itations in Section 7 and conclude with our future re-
search direction in Section 8.
96
Rizk, M., Fox, M. and Rosu, D.
Towards an Ontology of Task Dependence in Organizations.
DOI: 10.5220/0012238300003598
In Proceedings of the 15th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2023) - Volume 2: KEOD, pages 96-107
ISBN: 978-989-758-671-2; ISSN: 2184-3228
Copyright © 2023 by SCITEPRESS Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)
2 MOTIVATION
2.1 Case Study
The research discussed herein is based on a partner-
ship with the municipal government of one of the
ten largest cities in North America (henceforth re-
ferred to as “the City”). The objective of this on-
going partnership is to explore the potential for in-
formation technologies in aiding the City’s coordi-
nation of complex infrastructure projects. With the
City-led infrastructure projects growing larger, faster,
and more complex, effective coordination across the
City’s divisions, stakeholders, projects, and resources
has become critical for successful project delivery. A
case study of a major current infrastructure project
provided initial insights into the current coordination
practices and identified areas of potential improve-
ment using information technology.
The undertaking discussed in this section is con-
cerned with the City’s flood risk mitigation efforts
within one of its most flood-prone areas. The Riverine
Flooding Project (RFP), aimed at reducing riverine
flooding risks in the neighbourhood, involves design-
ing and implementing several pieces of infrastructure
such as bridges and concrete channels. Concurrent
with this, the City’s Urban Flooding Program (UFP)
is implementing sewage infrastructure improvements
that must be coordinated with the RFP. There are also
several planned capital works projects in the area that
the RFP must consider in its design. As operating in
this complex environment entails intense coordination
across various organizations, stakeholders, and tasks,
we had access to a wealth of coordination phenomena
and challenges to model and help address.
2.2 Challenges
While a detailed report on the case study will be
released separately, we outline our findings on the
nature of the coordination challenges for the pur-
pose of this paper. We identified three main types
of coordination challenges: participation, navigation,
and cross-cutting collaboration. Participation relates
to delays caused by inadequate stakeholder respon-
siveness. Navigation involves difficulties in identi-
fying affected stakeholders and accounting for their
needs. Cross-cutting collaboration is about synchro-
nizing the efforts of numerous stakeholders across
bureaucratic silos to meet interdependent objectives,
as opposed to independent pursuits that overlook the
broader project context.
Collectively, these challenges present consider-
able risks to the quality, budget, and timeline of in-
frastructure projects. Indeed, we identified cases of
delays, cost overruns, and suboptimal implementa-
tions that required rebuilding due to failures in partic-
ipation, navigation, and cross-cutting collaboration.
To successfully respond to these coordination
challenges, it is essential to identify the dependencies
necessitating coordination, the causes of coordination
failures, and strategies to mitigate the risk of future
failures. Therefore, an effective AI-enabled interven-
tion should be grounded in a robust representation of
dependence and the entities through which dependen-
cies materialize. In particular, we require the ability to
automatically infer instances of dependence, and cap-
ture the underlying causes and effects of dependen-
cies. As we will show in Section 3, the existing def-
initions of dependence and approaches to modelling
tasks are insufficient for addressing the observed co-
ordination challenges. We tackled these shortcomings
by making explicit the representational requirements
for capturing tasks and task dependencies and devel-
oping a new approach to modelling them, which we
introduce in the next section.
3 LITERATURE REVIEW
3.1 Ontological Analysis of Task
Dependence
This section describes the ontological analysis we
performed in order to provide a solid foundation for
the design and development of our framework. We
examined the organizational literature in pursuit of
understanding the underlying semantics of the con-
cepts of interest, and their relationships, as well as un-
covering any implicit assumptions or inconsistencies
that may be present. We established representational
requirements for our framework based on the previ-
ously outlined challenges and then generated a set
of dimensions that are relevant for effectively defin-
ing task dependence. These dimensions were sub-
sequently used to evaluate existing definitions and
implied conceptualizations in the existing literature,
and then propose our own definitions. With these in
hand, we reviewed the literature relevant to task mod-
elling (Section 3.2) in order to find frameworks that
could help operationalize our definitions and shape
our model.
3.1.1 Dimensions of Task Dependence
As touched upon in Section 2, a representation of the
dependencies that call for coordination is a prerequi-
site for any reasoning about the observed coordination
Towards an Ontology of Task Dependence in Organizations
97
challenges. Specifically, we need a model that repre-
sents the dependent entities (the subjects in need of
coordination), the situations that make them depen-
dent (the necessity for coordination), and the implica-
tions that the dependency has for them (the function
of coordination). From these reasoning tasks, a dis-
tinct set of representational requirements becomes ap-
parent for any model that we wish to utilize to aid our
use cases. More precisely, we need a model that cap-
tures the organization’s work, the dependency rela-
tionship’s basis between work, and the dependency’s
effects on the performance of work. Furthermore, we
need to capture each of these three requirements at
various levels of abstraction in the task structure, such
as the goal level and detailed activity level.
Our model must include a representation of the ac-
tivities to be performed and the desired states to be
achieved (i.e., goals). Therefore, our model must dis-
tinguish between states, the desire to achieve states,
the intention to work towards desired states, and the
specific activities performed to work towards desired
states. The “structure” of work, often referred to as
the task structure, must also be represented, which es-
sentially describes the decomposition of work.
To infer and model dependencies, we need to cap-
ture the fundamental “basis” of dependence between
two tasks. This can be seen as the core reason one task
may depend on another, or the irreducible essence of a
dependency relationship between two tasks (Herbsleb
and Roberts, 2006). To comprehend why a coordina-
tion failure may have occurred or to develop strategies
and mechanisms to mitigate them, we need to capture
the “effect” of a task dependency on the dependent
task. This effect indicates how a dependency influ-
ences task execution.
Moreover, to reason about task dependencies in
scenarios where the dependency’s exact nature cannot
be detailed, due to it being unknown or changing too
rapidly to warrant a specification, we need the defini-
tion of dependence to consider how it can be applied
to capture dependencies at various abstraction levels
of the work being analyzed. This highlights the dual
usage of the term “task” in the literature (achieve a
goal versus perform an activity), which we aim to dis-
ambiguate by this section’s end. For now, we refer to
this requirement as the “context” of a dependency.
Therefore, to meet the needs of our representa-
tional requirements, any approach for modelling task
dependence must explicitly (1) distinguish between
agent and task dependence; and allow for the (2) “ba-
sis”, (3) “effect” and (4) “context” of task dependence
to be made explicit. We will refer to these four condi-
tions in Section 3.1.5.
The underlying definitions for task and task struc-
ture must also be accounted for and must be con-
sistent, as these are the entities between which de-
pendencies exist, and the task structure serves as the
medium for these dependencies.
3.1.2 Definitions from Organizational Research
Multiple attempts to define task dependence exist in
the organizational literature. We focus our analysis
on four definitions due to their distinct perspectives,
allowing us to examine different elements of their def-
initions regarding the four dimensions we seek to ana-
lyze. For a more comprehensive review of task depen-
dence, we refer readers to (Raveendran et al., 2020).
Thompson (1967) provides one of the earliest def-
initions of task dependence, grounded in the relation-
ship between tasks’ inputs and outputs. He proposed
three types of possible task dependencies: pooled
(both tasks contribute to the same output), sequen-
tial (one task’s output is the input to another), and
reciprocal (both tasks feed into each other). Crow-
ston (1994), drawing from distributed artificial intel-
ligence literature, asserts that two tasks can only be
dependent via a common resource (which includes
states). He specifies three possible types of depen-
dencies: overlapping effects (two tasks contribute to
producing, or causing if the resource is a state, the
same resource); overlapping preconditions (two tasks
require the same resource); and overlapping effects
and preconditions (the resource produced by one task
is required by another). More recently, coming from
the emerging microstructural perspective, Puranam
(2018) suggests task interdependence is present when
the value generated from performing each task is dif-
ferent, depending on whether the other task is per-
formed or not. Finally, Raveendran et al. (2020) con-
sider task dependence’s role in informing organiza-
tion designs, depending on the nature of work being
well or ill-understood, further highlighting the context
of task dependence.
3.1.3 Analysis of Task
Prior to analyzing the four approaches’ interpretation
of task dependence, we need to grasp their fundamen-
tal assumptions about the nature of tasks and their
structure.
Thompson, while not explicitly defining a task,
implies that a task is a specific function or activity em-
bedded within a workflow, undertaken by agents with
defined inputs and outputs. The concept of “intention-
ality”, or the goal achieved by a task, is absent from
this definition. While intentionality may be argued to
be implicitly defined through the workflow in which
a given task is embedded, it may become increasingly
KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development
98
obscured as the task defined becomes more granular.
Crowston expands on this by considering tasks to
encompass both goal achievement and activity perfor-
mance. He characterizes goals as desired states and
activities as actions taken to achieve particular states.
However, Crowston’s harmonization of “achieving
goals” and “performing activities” under the single
concept of “task” is problematic. Goals and activ-
ities are inherently different; a goal is a desire to-
wards a state, while an activity is an action causing
a state. Moreover, Crowston’s confounding of re-
sources with states, and outcomes with outputs, fur-
ther limits the operationalization of this approach for
defining task dependence. To illustrate the limitations
of this approach, we can reference an upper-level on-
tology, such as DOLCE (Borgo et al., 2022). Re-
sources and outputs (resources that are produced by
an activity) are endurants while states and outcomes
(states caused by an activity) are perdurants.
Puranam provides a more explicit definition, de-
scribing a task as a transformation of inputs into
outputs in finite time, with an associated value de-
rived from the difference between inputs and outputs.
However, the relationship to a task’s corresponding
goals is not defined. Finally, Raveendran et al. fol-
low Puranam’s definition but separately consider goal
dependence. They focus on dependencies between
agents sharing common goals rather than dependen-
cies between goals themselves. This approach leaves
gaps in capturing task dependencies when the partic-
ular transformations for achieving these goals are not
clearly defined.
3.1.4 Analysis of Task Structure
In Thompson’s approach, the conceptualization of
task structure does not extend beyond the notion of
“workflow” and the categorization of activities into
input, technological, and output tasks.
Crowston’s view of task structure can be described
as a “means-ends” decomposition of intended “ef-
fects” (goals) into possible subgoals and primitive
activities. This approach attempts to streamline the
analysis of tasks by harmonizing goals and activities.
However, the confounding of states, intentions, and
actions confuses the definition, preventing a clear and
unambiguous formal representation of task structure.
Puranam defines a task structure as the most fine-
grained means-end decomposition of an organiza-
tion’s goals into its constituent tasks and their inter-
dependencies, operationalized using design structure
matrices (Baldwin and Clark, 2000). However, this
approach raises a couple of issues. Firstly, the ab-
sence of explicit representation of goals, intermedi-
ate goals, and their relationship to activities is prob-
lematic. Unlike Crowston, who incorporated the de-
composition of goals into actions in his notion of task
structure, Puranam’s approach lacks this dimension
explicitly. Instead, decomposition is implicitly de-
fined through the clustering of interdependent tasks.
This absence hinders us from modelling the connec-
tion between a given goal and the actions required
to realize it, largely limiting our ability to precisely
represent the decomposition of high-level tasks into
their constituent sub-tasks and corresponding sub-
states they are intended to achieve. Secondly, the
modelling of decomposition is paradoxically bottom-
up rather than top-down. Task structure is defined
based on task dependencies, rather than defining task
dependence based on task structure. This can be lim-
iting when our goal is to infer dependencies between
tasks that are hard to detect, given a representation of
the tasks.
The limitations associated with Puranam’s ap-
proach to modelling task structure are also applicable
to Raveendran et al.s approach. They suggest that the
level of understanding of a work environment can be
expressed in terms of the extent to which the underly-
ing task structure can be modelled in terms of detailed
tasks. However, this approach does not solve the is-
sues raised in Puranam’s, maintaining the difficulties
for our purpose of inferring obscure dependencies.
3.1.5 Analysis of Task Dependence
In this section, we explore how each of the four
task dependence approaches meets the four condi-
tions specified at the end of Section 3.1.1.
Separation from Agent Dependence. Thompson
defines dependence between workflows across orga-
nizational units, implying task dependence is agent
dependence caused by the relationship between the
tasks of the dependent agents. Conversely, Crowston
and Puranam assert task dependence as separate from
agents. Furthermore, Puranam shows that task depen-
dence is neither a necessary nor sufficient condition
for agents of interdependent tasks to be interdepen-
dent, challenging Thompson’s perspective. Instead,
the relationship between task and agent dependence is
influenced by the extent to which the tasks are under-
stood, as noted by Raveendran et al. (2020). In well-
understood environments, agent dependencies stem
from task dependence, through task allocation deci-
sions. In less clear environments, task dependence
emerges from the collective sense-making processes
shaped by agent dependence.
Basis of Task Dependence. Thompson views task
dependence as arising from workflow directionality,
which is defined in terms of overlaps between the in-
puts and outputs of tasks. In Crowston’s approach, the
Towards an Ontology of Task Dependence in Organizations
99
basis of dependence is the common resource through
which two tasks have an overlap in their precondi-
tions and/or effects. Both approaches have limita-
tions. Thompson’s approach overlooks dependencies
based on information flows that are not necessarily
isomorphic with workflow directionality. Crowston’s
approach, however, captures some of these dependen-
cies, though not without its own limitations, as will be
shown shortly. For instance, Malone and Crowston
(1994), show that a producer-consumer relationship
between two tasks (i.e., the output of one task is the
input of another) may be due to either an inventory
(the second task can only execute after the output of
the first is available) or usability constraint (the first
task must produce its output in a way that is usable
by the second task). The former constraint, a sequen-
tial dependency in Thompson’s language, matches the
workflow direction. However, the latter constraint,
grounded in information flow, runs in the opposite di-
rection to workflow and would not be definable with
Thompson’s approach.
While broader than Thompson’s basis, a limita-
tion of Crowston’s approach is that it is grounded in
whether a common resource exists between two tasks,
without an explicit representation of the underlying
reason why the common resource forms a depen-
dency. For instance, the distinction between inven-
tory and usability constraints is left to a human rea-
soner applying the model. This distinction is impor-
tant since the reason why a common resource causes a
dependency has implications for how the dependency
ought to be managed, as we’ll see shortly when dis-
cussing “effects”.
Puranam and Raveendran et al. propose a differ-
ent perspective, focusing on the mutual influence be-
tween tasks on the value they generate as the basis of
dependence. However, the basis here is overly con-
strained as it only considers if the value of one task
varies with if another task is performed, rather than
how it is performed. Additionally, though “value” is
intended to be an abstract construct which can be arbi-
trarily defined, the approach does not allow for an ex-
plicit representation of why a differential in the value
of a task exists in the first place.
Effect of Task Dependence. Thompson’s ap-
proach to managing task dependencies involves var-
ious coordination mechanisms that vary in “strength”
based on the type of task dependence. This suggests
effects of task dependencies include interaction inten-
sity and complexity. However, Thompson’s frame-
work does not capture the precise ways in which a
task can be constrained by dependencies.
Crowston and Puranam also touch on “effects”
but don’t explicitly address how tasks may vary in
their performance due to dependencies. For each type
of dependency, Malone and Crowston (1994) offer a
set of coordination mechanisms for managing the de-
pendency, but they do not explicitly model the pre-
cise way in which the coordination mechanisms af-
fect how tasks are performed (i.e., why the coordina-
tion mechanisms work). For Puranam’s approach, the
specific dimensions of a task’s performance that may
change due to its value relation to another task are
not explicitly defined. Without a representation of the
“effects” of dependence on the performance of a task,
we are limited in our ability to reason about the effi-
cacy of alternative coordination strategies for a given
dependency.
Context of Task Dependence. Given that the
approaches offered by Thompson, Puranam, and
Raveendran et al. view tasks as well-defined opera-
tions, they are limited in their ability to capture task
dependencies in environments where the nature of
work is not well understood and tasks cannot be spec-
ified beyond the goals they are intended to achieve
(i.e., the specific activities are either unknown or can-
not be explicitly represented). Crowston’s definition
of tasks as both the achievement of goals and the
performance of activities allows for the capturing of
dependencies in ill-understood work environments,
where only dependencies between the achievement
of goals can be represented. However, due to the
limitations specified in Section 3.1.3, operationaliz-
ing Crowston’s approach is a challenge.
3.1.6 Proposed Definitions
Based on the ontological analysis of tasks, task struc-
ture, and task dependence, we have demonstrated that
the approaches offered by the organizational literature
are neither consistent with one another, nor do they
adequately satisfy the conditions that we require to
reason about addressing the coordination challenges
we found in practice. Below, based on our ontologi-
cal analysis and required conditions, we propose a set
of definitions that form the basis of our model:
Task: The intention of an agent(s) to work to-
wards some goal. A well-understood task can be
further specified in terms of activities that when
executed cause the desired states, whereas an ill-
understood task can only be specified in terms of
the goal that is intended to be accomplished.
Activity: An activity can be performed by an
agent, and may have inputs (required resources)
and outputs (produced resources). An activity
causes outcomes, which are the consequences of
an activity’s performance. Activities may have
characteristics which define particular ways in
KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development
100
which the performance of the activity can vary.
Task Structure: A task structure is a decompo-
sition of tasks into the constituent activities that
must be performed, and the corresponding states
that must hold in order for the goal of the task
to be achieved. Activities that cause states may
be decomposed into subactivities that cause sub-
states, and so forth. In essence, a task structure
specifies “how” a task to bring about some goal
is achieved in terms of the activities that must be
performed and the states that must be true.
Task and Activity Dependence: Task A is de-
pendent on Task B if the way in which the goal of
A is achieved is constrained by the way in which
the goal of B is achieved. Activity A is dependent
on Activity B if the way in which A is performed
is constrained by the performance of B.
Here, we explicitly capture that the basis of the
dependency is that there are constraints placed on the
performance of activity A by the performance of B.
Additionally, the concept of “value” is implicitly cap-
tured since it can be assumed that the basis of con-
straint is that in order for the activity to be performed
in a way that maximizes its value, the particular way
in which it should (or must) be performed may vary
with how another activity is performed (rather than
just if ). More so, the distinction between task and ac-
tivity dependence allows us to capture dependencies
between work in both well- and ill-understood envi-
ronments.
3.2 Related Task Frameworks
The concept of a task has been extensively examined
across various domains, ranging from psychology
(Annett and Duncan, 1967) to organizational stud-
ies and information and computer science (Vernadat,
2020; Alam et al., 2015; Guizzardi et al., 2013; Yu
and Mylopoulos, 1995; Greenspan et al., 1994), pre-
senting an array of perspectives and methodologies.
These span from cognitive theory-based methods fo-
cusing on human thought and behaviour, to methods
designed to endow robots with autonomy.
An early, and subsequently prominent task mod-
elling framework, the Hierarchical Task Analysis
(HTA) (Annett and Duncan, 1967) supported the view
that a task is “any piece of work that has to be done”
and that task performance “is a goal-directed be-
haviour”. Their approach, departing from earlier ap-
proaches, was to offer functional analysis rather than
behavioural descriptions of tasks, initiating the anal-
ysis from the task’s goals rather than associated ac-
tivities. However, despite its name, HTA, does not,
in fact, emphasize tasks, but focuses instead on goals
and the operations to achieve them. Due to its onto-
logical commitments, this framework, and its subse-
quent extensions, do not provide the means to capture
and distinguish an activity from the intent to perform
it, which is an essential requirement that our represen-
tational framework strives to meet.
Successive approaches (Phipps et al., 2011; Stan-
ton, 2006) emerged in organizational sciences as well
as computer and information science. Of particular
relevance to our work are the frameworks developed
from the field of enterprise modelling, which sits at
the intersection of IT, information science, and orga-
nization studies. Also relevant, from the computer
science literature, are frameworks emerging from the
fields of requirements engineering (e.g., i*, KAOS)
and AI planning.
Enterprise modelling, defined by Fox and
Gr
¨
uninger (1998) as “a computational representation
of the structure, activities, processes, information re-
sources, people, behaviour, goals and constraints of a
business, government, or other enterprise”, has gen-
erated various complementary frameworks over four
decades. These are centred around activities (e.g.,
IDEF, MERISE), business processes (e.g., ARIS,
CIMOSA), and enterprise knowledge (e.g., DEMO
by Dietz,1999; and TOVE by Fox et al.,1995). While
these frameworks incorporate elements of agents,
goals, tasks, and activities, none distinguish between
an activity and the intent to execute it.
AI planning shares similar concerns with our cur-
rent work, striving for computationally executable
plans along with methodologies for social and cog-
nitive plans. Existing methodologies (e.g., Chan-
drasekaran and Josephson, 1997; and Bermejo-
Alonso, 2018) do attempt to capture the activity to
perform and how it should be executed as part of a
plan. However, they do not account for the intent to
perform an activity, a crucial distinction from the goal
the activity is designed or expected to achieve.
There are also domain-independent efforts, such
as upper-level ontology work (e.g., BFO
1
, UFO
2
,
DOLCE
3
, DOLCE-Lite-Plus
4
) that offer a mecha-
nism for modelling agents, goals, tasks, and actions.
DOLCE-Lite-Plus, in particular, defines a task as a
“course used to sequence activities or other control-
lable perdurants”, where a course is a “concept that
selects (in particular, it sequences) perdurants (pro-
cesses, events, or states), as a component of some s-
1
https://basic-formal-ontology.org/
2
https://nemo.inf.ufes.br/en/projetos/ufo/
3
http://www.loa.istc.cnr.it/dolce/overview.html
4
https://www.w3.org/2001/sw/BestPractices/WNET/
DLP3941 daml.html
Towards an Ontology of Task Dependence in Organizations
101
description” (which relates, roles, endurants and con-
texts). In effect, tasks are (the desired) targets of some
role played by an agent and can relate to ground ac-
tivities or decision-making. In this framework, tasks
are explicitly declared to be disjoint from actions and
action types, an ontological stance that is compatible
with our representational requirements.
4 CONCEPTUAL MODEL
In this section, we introduce our conceptual model,
with each subsection covering a different layer. Fig-
ure 1 displays the ontology design pattern, highlight-
ing the main concepts and relations discussed.
4.1 Agent, Intentions, and Desires
The foundational concepts in our model are agents,
their tasks, and their goals. We define an Agent as
an entity capable of possessing goals, having inten-
tions to strive for those goals, and executing actions
to attain them. Agents can be individuals, teams, de-
partments, or entire organizations and even temporar-
ily formed entities like committees or task forces. A
Task signifies an agent’s intention to exert effort to-
wards a goal. It is defined in terms of the agent with
the intent and the goal, which is the target of this in-
tent. A Goal represents a desired “state of affairs”.
It is defined in terms of the agent who has the goal
and the desired State (detailed later) of the agent. At
this stage, we don’t consider the reasons behind an
agent’s goal or task. Also worth noting, our model
doesn’t incorporate organizational roles since our fo-
cus is on capturing dependencies between work, not
dependent on specific organizational settings or for-
mal structures.
Tasks may also be defined via Activities through
which goals can be achieved. The difference between
tasks and activities is important yet subtle. Tasks con-
vey the intent to work towards a goal, while activities
are concrete operations that might lead to a state, re-
gardless of whether that state is desirable. Tasks rep-
resent the intention to act towards a goal, while activ-
ities specify the exact manner in which one can act.
This distinction is useful in contrasting well-
defined and ill-defined task structures. For well-
defined task structures, where necessary actions to
achieve a goal are known, we can understand the spe-
cific contribution of each agent. Conversely, in ill-
defined task structures, where agents’ specific actions
are unpredictable or can’t be predefined, we can still
capture their collective intention towards the shared
goal. Furthermore, this distinction allows us to ex-
press task dependence in terms of achieving goals
rather than merely performing activities. In scenar-
ios where the accomplishment of one goal influences
or restricts another, but the exact method of influence
can’t be predetermined (as in ill-defined task struc-
tures), we model tasks as the intention to work to-
wards a goal. This way, we can articulate the influ-
ence of one goal on another through the tasks involved
in achieving both goals. Thus, the achievement of
a state may be constrained by how another state is
achieved, even if the precise “how” (i.e., an activity)
is not known or cannot be represented. This gives us
a higher level of representation, at the level of “inten-
tion to act towards” rather than the specific “act”.
4.2 Activities and Resources
An Activity refers to a well-defined operation that an
agent can perform as part of a task, causing an out-
come state. It may also be enabled by a state and may
require or produce a Resource. An activity can be
seen as a production technology (Puranam, 2018) or
an IDEF
/
0 function. An activity may have an Activ-
ity Decomposition, which breaks down complex ac-
tivities into constituent sub-activities. This construct
captures relationships between activities at different
abstraction levels, reminiscent of hierarchical mod-
elling in IDEF
/
0 or the AggregateActivity construct in
TOVE (Fox et al., 1993). Explicit modelling of the
decomposition as a construct allows representation,
querying, and visualization of task structures at var-
ious abstraction levels. Capturing the specific ways
in which an activity’s performance can vary is crucial
for activity dependence (the performance of one ac-
tivity being influenced by another). For this, we use
the Activity Characteristic construct, defining a feature
of an activity subject to variation. Not every activity
characteristic may be part of a dependency. We clas-
sify these characteristics into five main categories (not
intended to be collectively exhaustive):
Input characteristics: Variations in activity inputs
such as material selection, resource consumption,
cost, and information source.
Process characteristics: Variations in the method
or implementation of an activity.
Output characteristics: Activity performance di-
mensions that define the output, like quality,
quantity, and design.
Spatial characteristics: Activity performance di-
mensions related to the physical or virtual loca-
tion where an activity occurs.
Temporal characteristics: Temporal features such
as start time, end time, and duration.
KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development
102
Figure 1: Task dependence ontology design pattern showing main classes and properties (dashed arrows for subclassOf).
4.3 States
A State describes a particular aspect of an object or
situation, capturing the idea that the modelled object
or situation can have varying conditions over time.
States can be complex or atomic. A Complex State
can be either a Conjunctive State or a Disjunctive
State, representing the conjunction or disjunction of
other (sub)states, respectively.
An Atomic State is defined in terms of a state
characteristic, operator, and value combination. The
State Characteristic is an identifiable property of an
object or situation. The Operator defines the relation-
ship between a characteristic and value (e.g. , ,
=, ̸=). The Value represents the specific value of the
state characteristic defined in an atomic state. It can
be based on a unit of measure, an ordinal or nom-
inal scale, or any data type. This flexibility allows
for the representation of specific states and reason-
ing about the achievement of goals that can be sat-
isfied as well as satisficed (Simon, 1947), a feature
also seen in i* (Yu and Mylopoulos, 1995). This is
useful when the exact condition for state satisfaction
cannot be predetermined (e.g., minimize construction
costs) or when a particular state characteristic is hard
to quantify (e.g., improve design aesthetic).
4.4 Dependence
A Basis of Dependence captures the underlying rea-
son for one task, activity, or activity characteristic de-
pending on another, establishing the nature of their
relationship. This construct is key to our model, as
it underpins how task and activity dependencies are
inferred. A basis of dependence may constrain or be
altered by how a task is carried out, an activity is per-
formed, or an activity characteristic varies. If a task or
activity (or one of its characteristics) is constrained by
a basis of dependence and that basis is altered by an-
other task or activity, then the former entity depends
on the latter. The semantics of the depends on rela-
tion vary based on the class of entities being related.
Also crucial to defining a basis of dependence is
its dependum – the entity through which a basis of de-
pendence exists. For example, if an activity produces
a resource used by another, then the availability of the
resource is the basis of dependence and the resource
is the dependum. While this is not an exhaustive clas-
sification of all possible bases of dependence, we’ve
identified six that collectively allow the representation
of a variety of dependencies between organizational
tasks and activities:
Availability: The availability of a common re-
source (either as a common input or intermedi-
ary object) may be the basis of dependence be-
tween two tasks or activities. Here, the common
resource is the dependum.
Functional: Arises when the ability to perform
an activity (both dependum and subject of depen-
dence) varies with if or how another activity is
performed or task accomplished.
Complementarity: This basis of dependence
comes into play when the overall effects of two
activities or tasks are either greater or less than
the sum of their parts. The non-additive state char-
acteristic is the dependum here. For example, to
minimize traffic disruptions, a municipality may
seek to bundle several planned maintenance activ-
ities (e.g.; road repavement, sewer improvements,
utility relocation) to occur at the same time.
Towards an Ontology of Task Dependence in Organizations
103
Compatibility: This basis of dependence arises
when two tasks or activities need to coexist, due
to some state or activity characteristic condition,
but may not functionally influence each other.
Uncertainty: This basis of dependence comes into
play when two tasks or activities depend on each
other due to the uncertainty of a state or activity
characteristic.
Complexity: This basis of dependence applies
when the exact nature of the dependency relation-
ship between two tasks or activities is unknown or
cannot be explicitly stated.
We don’t impose any set relations between these
bases of dependence; depending on the domain the
ontology is applied to, some bases may be sub-bases
of others or disjoint. The complexity basis may serve
as a catch-all for bases of dependence that can’t be
captured by the ones we’ve defined.
To illustrate the model, consider two activities
requiring the same non-shareable, non-consumable,
unary-capacity resource. An availability basis of de-
pendence exists between them through the shared re-
source. Since the start time of either activity affects
the resource’s availability, both activity characteris-
tics “alter” the basis of dependence. As the resource
availability constrains when each activity can start,
the basis of dependence “constrains” both activities.
Hence, they both depend on each other.
5 FORMALIZATION AND
IMPLEMENTATION
In this section, we provide a brief snapshot of
our ontology-based formalization of the conceptual
model, formulated in first-order logic (FOL). Cho-
sen for its expressivity and the ability to capture ad-
vanced rules, FOL equips us with the complex rea-
soning capabilities necessary for our use cases, such
as inferring instances of dependence. Despite its po-
tential lack of decidability that might make alterna-
tives like description logic appealing, we opted to first
construct our model “as intended” before translating
into a description logic, avoiding upfront constraints
on expressivity and inference.
Space constraints necessitate us to only include in
this paper the axiomatization of a subset of the depen-
dency inferences between tasks, activities, and activ-
ity characteristics. Other constructs underpinning ax-
ioms (1) to (5) are summarized in Table 1. The current
version of the full axiomatization can be found in the
Task Dependence Ontology Github repository
5
. The
FOL specification was implemented in Prover9 and
verified for consistency. Additionally, we developed
a consistency-verified OWL (and SWRL) implemen-
tation to allow for further testing and future integra-
tion of the ontology into decision-support tools. Both
implementations can be found in the repository.
An activity characteristic dependsOn another ac-
tivity characteristic if there is a basis of dependence
that constrains the former and is also altered by the
latter:
b, a
1
, a
2
, c
1
, c
2
(constrains(b, c
2
)
alteredBy(b, c
1
) (c
1
̸= c
2
)
hasActivityCharacteristic(a
1
, c
1
)
hasActivityCharacteristic(a
2
, c
2
)
dependsOn(c
2
, c
1
))
(1)
If an activity characteristic dependsOn another activ-
ity characteristic, then the activity of the former char-
acteristic is dependent on the activity of the second
characteristic:
a
1
, a
2
, c
1
, c
2
(Activity(a
1
) Activity(a
2
)
hasActivityCharacteristic(a
1
, c
1
)
hasActivityCharacteristic(a
2
, c
2
)
dependsOn(c
2
, c
1
) (a
1
̸= a
2
)
dependsOn(a
2
, a
1
))
(2)
If an activity dependsOn another activity, then the
task that the former activity is a part of is dependent
on the task that the second activity is a part of:
t
1
, t
2
, a
1
, a
2
(Task(t
1
) Task(t
2
)
hasActivity(t
1
, a
1
) hasActivity(t
2
, a
2
)
dependsOn(a
2
, a
1
) (t
1
̸= t
2
)
dependsOn(t
2
, t
1
))
(3)
An activity dependsOn another activity if there is a
basis of dependence that constrains the former and is
also altered by the latter:
a
1
, a
2
, b(Activity(a
1
) Activity(a
2
)
constrains(b, a
1
) alteredBy(b, a
2
)
(a
1
̸= a
2
) dependsOn(a
1
, a
2
))
(4)
A task dependsOn another task if there is a basis of
dependence that constrains the former and is also al-
tered by the latter:
t
1
, t
2
, b(Task(t
1
) Task(t
2
) constrains(b, t
1
)
alteredBy(b, t
2
) (t
1
̸= t
2
) dependsOn(t
1
, t
2
))
(5)
6 APPLICATION
In this section, we describe our framework’s appli-
cation to the project introduced in Section 2 to val-
idate its representational effectiveness (visualized in
5
https://github.com/rizkmena/Task-Dependence-Ontology
KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development
104
Table 1: Summary of axiomatization of the task dependence ontology’s core concepts.
Concept Summary of Axioms
Agent An entity that has a hasTask relation to a Task and a hasGoal relation to a Goal.
Task Has a taskOf relation with the Agent that has the Task, and the Goal that the Task is towards. May have Activity(s) associated with
it via a hasActivity relation.
Goal Defined in terms of a goalOf relation with the Agent that has the Goal, and a hasDesiredState relation with the State that the goal
is about.
Activity Can be performedBy an Agent, is an activityOf a Task, and causes a State. Note that the State that is causedBy an Activity is an
Outcome. An activity may also be enabledBy a State.
ActivityDecomposition Is a decompositionOf an Activity and a compositionOf at least two other Activity entities (which are the subactivities of the decom-
posed activity).
ActivityCharacteristic An Activity may have a hasActivityCharacteristic relation with an ActivityCharacteristic. TemporalCharachterstic, SpacialCharac-
teristic, InputCharacteristic, ProcessCharacteristic, and OutputCharacteristic are types of ActivityCharacteristic.
Resource An entity that is either requiredBy or producedBy an Activity. Can be a Shareable or NonShareableResource, and a Consumable or
NonConsumeableResource.
State A State may either be a ComplexState or AtomicState. A ComplexState has at least one hasSubState relation with another State, and
may be either a ConjunctiveState or a DisjunctiveState. An AtomicState has a hasStateCharacteristic, hasOperator, and hasValue
relations with a StateCharacteristic, Operator, and Value, respectively.
BasisOfDependence Has a hasDependum relation with either a Task, Activity, ActivityCharacteristic, Resource, or StateCharacteristic. Can have con-
strains and alteredBy relations to a Task, Activity, or ActivityCharacteristic. Availability, Complementarity, Functionality, Compat-
ibility, Uncertainty, and Complexity are types of BasisOfDependence.
Figure 2) and demonstrate the type of insights it can
generate. Prover9 and OWL implementations of this
section are also provided in the Github repository, to
allow for independent verification that our framework
supports generating the inferences described in this
section. Given space constraints, we focus on the ele-
ments that illustrate dependencies, rather than a com-
prehensive task structure. We will explore three in-
stances of dependencies, each at different abstraction
levels, to demonstrate the model’s ability to identify
dependencies at both well- and ill-defined task struc-
ture levels.
The agents are the Riverine Flood Risk Mitiga-
tion Team (RM), and the City’s Water Infrastruc-
ture Management (WIM) and Transportation Services
(TS) departments. The neighbourhood of interest is
grappling with substantial flood risks due to both ur-
ban (sewer overflows) and riverine (watercourse over-
flows) sources. RM is designing a bridge as part of its
Riverine Flooding Project (RFP). WIM must redesign
a sanitary trunk sewer (STS) for capacity expansion
and execute the Urban Flooding Program (UFP) to
reduce urban flooding. For the STS, WIM must de-
sign a new storage pipe and require road access for
UFP sewer improvements. TS must repave a road to
maintain good road conditions. Three dependencies
inferred here are:
Dependency 1. Flooding is a complex issue due
to riverine and urban flooding being interconnected
sources. If sewer systems are improved to reduce
urban flooding, but the region remains vulnerable to
a nearby overflow-prone watercourse (riverine risk),
flooding will still occur. The flood protection level
in a given area is only as strong as the weakest link.
Hence, there is a complementary basis of dependence
among tasks to reduce various types of flood risk. Par-
ticularly, executing the UFP depends on the STS and
riverine developments, since the specific sewer im-
provements are constrained by the existing flood risk
altered by these two developments.
Dependency 2. Co-location demands compatibil-
ity between the design of the RFP bridge and the
STS’s storage pipe. Although they don’t impact
each other’s functionality, their coexistence estab-
lishes a compatibility-dependent relationship. Since
the bridge design precedes the storage pipe design, it
alters this dependency and constrains the pipe design,
which must align with the bridge design.
Dependency 3. TS’s road paving and WIM’s sewer
improvement activities require access to the same
road, causing an availability-based dependency. The
start times of each activity alter and are constrained
by the availability, creating mutual dependence.
Based on the representation of these dependen-
cies, we are afforded the ability to reason about how
to address them. The UFP execution’s dependency
on the RFP and STS developments informs the con-
siderations WIM must make. For instance, post-
development hydrological models could guide the
UFP’s activities. The dependency between Activity 1
and Activity 2 necessitates information exchange on
the bridge design to facilitate a compatible pipe de-
sign. With Activity 3 and Activity 4 mutually depen-
dent due to timing, we see that coordination in terms
of scheduling is necessary, positioning a schedule as
a coordination mechanism. This awareness can also
help WIM and TS to synergistically overlap their ac-
tivities, reducing public disruptions and setup costs.
In summary, the model effectively encapsulates
the task environment at varied specificity levels, iden-
tifies task and activity dependencies, uncovers depen-
dency roots, and suggests coordination methods.
7 DISCUSSION
In our discussion, we draw attention to three key, mu-
tually reinforcing benefits of our framework: multi-
level abstraction, inferred dependencies, and en-
Towards an Ontology of Task Dependence in Organizations
105
Figure 2: Partial representation of three case study scenarios. Pink relations indicate dependencies that can be inferred.
hanced reasoning about coordination. The precise
semantics of our axiomatization, linking tasks and
activities through bases of dependence, allows auto-
mated identification of critical and potentially latent
dependencies. By facilitating the discovery of depen-
dencies that might otherwise go unnoticed, coordina-
tion failures can be preempted and opportunities for
improved integration can be revealed, as exemplified
by the third dependency outlined in Section 6.
As observed in our case study, many project ele-
ments required rework due to a lack of dependency
awareness. This could be avoided by taking advan-
tage at project design time of decision-support tools
that are able to alert decision-makers of potential in-
fluences on or from the various project tasks. Our for-
mally axiomatized and implemented framework can
serve as the backbone of such tools.
Secondly, by distinguishing the intent to achieve
goals from the activities performed to reach them, the
model captures dependencies at various task abstrac-
tion levels, making it applicable in both well- and ill-
understood task environments. This makes our frame-
work practically useful for managing dependencies
across strategic, tactical, and operational tiers of pol-
icy, program, and project management when embed-
ded in a decision-support tool.
Finally, by expressing dependence through its ba-
sis and effect, our model facilitates a deeper under-
standing of coordination failures’ root causes and re-
veals potential mitigation strategies. It offers a new
perspective on coordination failures by treating them
as unsatisfied (or unoptimized) constraints (Herbsleb
and Roberts, 2006), with the basis of dependence
representing the unsatisfied constraint and the effect
revealing the “violating value”. Consequently, our
model can be viewed as a higher-level abstraction
of constraint networks, providing an explicit frame-
work for reasoning about strategies to “solve” con-
straint networks based on various activity character-
istics. For instance, while the set of dependsOn re-
lations between temporal activity characteristics can
be seen as forming a temporal constraint network, for
which several algorithms exist (Dechter et al., 1991),
we can now model constraint networks based on other
activity attributes, such as cost, design, and quality
networks. With this perspective, we can work towards
both human- and computer-based coordination strate-
gies for “solving” the various kinds of networks.
However, our framework, in its current form, also
has limitations. Primarily, the bases of dependence
are asserted, not inferred. Ideally, the existence of
all bases of dependence would be automatically in-
ferred based on a robust representation of task struc-
tures. For instance, while availability can be in-
ferred from the current axiomatization, complemen-
tarity cannot be, due to the present inability to repre-
sent “non-additivity” in state characteristics. There-
fore, a more detailed representation of activities and
states is needed to facilitate the automated detection
of bases of dependence. Moreover, the framework
does not support accounting for the relative impor-
tance of the various dependencies impacting a task or
activity, thus failing to explicitly capture the signifi-
cance of a constraint amidst large networks of depen-
dencies with inherent trade-offs. These aspects call
for refinement of our model in future work.
8 CONCLUSION
In this paper, we present a novel approach to repre-
senting tasks, task structures, and task dependencies
to help enhance coordination in organizations. Faced
with the coordination issues identified in real-life case
KEOD 2023 - 15th International Conference on Knowledge Engineering and Ontology Development
106
studies shared by domain experts, we examined the
organizational and enterprise modelling literature for
applicable modelling frameworks. We concluded they
did not meet the representational requirements we in-
ferred from our use cases. As such, we developed a
new model of task dependence, grounded in the no-
tion of constraint, and underpinned by two new con-
structs that define the basis and effects of a depen-
dency. We discussed a practical application of our
framework to a real-world scenario and showed how
it enables the inference of (latent) dependencies and
the reasoning about coordination strategies to address
them. In future work, we plan to extend the frame-
work and its ability to support organization design de-
cisions and facilitate the integration of work by tack-
ling the modelling of inter-agent dependencies.
REFERENCES
Alam, K. A., Ahmad, R., Akhunzada, A., Nasir, M. H.
N. M., and Khan, S. U. (2015). Impact analysis and
change propagation in service-oriented enterprises: A
systematic review. Information Systems, 54:43–73.
Annett, J. and Duncan, K. D. (1967). Task analysis and
training design. Occupational Psychology, 42:211–
221.
Baldwin, C. Y. and Clark, K. B. (2000). Design Rules:
The Power of Modularity. The MIT Press, Cambridge,
Mass.
Bermejo-Alonso, J. (2018). Reviewing task and planning
ontologies: An ontology engineering process. In
Aveiro, D., Dietz, J. L. G., and Filipe, J., editors,
Proceedings of the 10th International Joint Confer-
ence on Knowledge Discovery, Knowledge Engineer-
ing and Knowledge Management, IC3K 2018, Vol-
ume 2: KEOD, Seville, Spain, September 18-20, 2018,
pages 181–188.
Borgo, S., Ferrario, R., Gangemi, A., Guarino, N., Ma-
solo, C., Porello, D., Sanfilippo, E. M., and Vieu,
L. (2022). DOLCE: A descriptive ontology for lin-
guistic and cognitive engineering. Applied Ontology,
17(1):45–69.
Chandrasekaran, B. and Josephson, J. R. (1997). The ontol-
ogy of tasks and methods.
Crowston, K. (1994). A taxonomy of organizational de-
pendencies and coordination mechanisms. Technical
Report 174, Massachusetts Institute of Technology.
Dechter, R., Meiri, I., and Pearl, J. (1991). Temporal con-
straint networks. Artificial intelligence, 49(1-3):61–
95.
Dietz, J. L. (1999). Understanding and modelling busi-
ness processes with DEMO. In Conceptual Model-
ing—ER’99: 18th International Conference on Con-
ceptual Modeling Paris, France, November 15–18,
1999 Proceedings 18, pages 188–202.
Fox, M., Barbuceanu, M., and Gruninger, M. (1995). An
organisation ontology for enterprise modelling: pre-
liminary concepts for linking structure and behaviour.
In Proceedings 4th IEEE Workshop on Enabling Tech-
nologies: Infrastructure for Collaborative Enterprises
(WET ICE ’95), pages 71–81.
Fox, M. S., Chionglo, J. F., and Fadel, F. G. (1993). A
common-sense model of the enterprise. In Proceed-
ings of Industrial Engineering Research Conference,
pages 178–194.
Fox, M. S. and Gr
¨
uninger, M. (1998). Enterprise modeling.
AI Mag., 19:109–121.
Greenspan, S., Mylopoulos, J., and Borgida, A. (1994).
On formal requirements modeling languages: RML
revisited. In Proceedings of the 16th International
Conference on Software Engineering, ICSE ’94, page
135–147, Washington, DC, USA. IEEE Computer So-
ciety Press.
Guizzardi, R. S. S., Franch, X., Guizzardi, G., and
Wieringa, R. J. (2013). Ontological distinctions be-
tween means-end and contribution links in the i*
framework. In Ng, W., Storey, V. C., and Trujillo,
J., editors, Conceptual Modeling - 32th International
Conference, ER 2013, Hong-Kong, China, November
11-13, 2013. Proceedings, volume 8217 of Lecture
Notes in Computer Science, pages 463–470.
Herbsleb, J. D. and Roberts, J. A. (2006). Collaboration in
software engineering projects: A theory of coordina-
tion. In Proceedings of the International Conference
on Information Systems, pages 553–568.
Malone, T. W. and Crowston, K. (1994). The interdisci-
plinary study of coordination. ACM Computing Sur-
veys, 26(1):87–119.
Phipps, D. L., Meakin, G. H., and Beatty, P. C. (2011). Ex-
tending hierarchical task analysis to identify cognitive
demands and information design requirements. Ap-
plied Ergonomics, 42(5):741–748.
Puranam, P. (2018). The Microstructure of Organizations.
Oxford University Press, Oxford, United Kingdom.
Puranam, P., Raveendran, M., and Knudsen, T. (2012).
Organization Design: The Epistemic Interdepen-
dence Perspective. Academy of Management Review,
37(3):419–440.
Raveendran, M., Silvestri, L., and Gulati, R. (2020). The
Role of Interdependence in the Micro-Foundations of
Organization Design: Task, Goal, and Knowledge
Interdependence. Academy of Management Annals,
14(2):828–868.
Simon, H. A. (1947). Administrative Behavior: A Study of
Decision-Making Processes in Administrative Orga-
nizations. Macmillan, New York, NY.
Stanton, N. A. (2006). Hierarchical task analysis: Devel-
opments, applications, and extensions. Applied Er-
gonomics, 37(1):55–79. Special Issue: Fundamental
Reviews.
Thompson, J. D. (1967). Organizations in action: Social
science bases of administrative theory. McGraw-Hill,
New York, NY.
Vernadat, F. (2020). Enterprise modelling: Research review
and outlook. Computers in Industry, 122:103265.
Yu, E. S. and Mylopoulos, J. (1995). From E-R to “A-
R”—modelling strategic actor relationships for busi-
ness process reengineering. International Journal of
Cooperative Information Systems, 4(2):125–144.
Towards an Ontology of Task Dependence in Organizations
107