ITM Web of Conferences 15, 01002 (2017)

DOI: 10.1051/itmconf/20171501002
CMES’17

Decomposition of business process models into reusable

sub-diagrams
Piotr Wiśniewski1,*
1
AGH University of Science and Technology, Department of Applied Computer Science, al. Mickiewicza 30, 30-059 Krakow, Poland

Abstract. In this paper, an approach to automatic decomposition of business process models is proposed.
According to our method, an existing BPMN diagram is disassembled into reusable parts containing the
desired number of elements. Such elements and structure can work as design patterns and be validated by
a user in terms of correctness. In the next step, these component models are categorised considering their
parameters such as resources used, as well as input and output data. The classified components may be
considered a repository of reusable parts, that can be further applied in the design of new models. The
proposed technique may play a significant role in facilitating the business process redesign procedure,
which is of a great importance regarding engineering and industrial applications.

1 Introduction representation. The proposed decomposition algorithm is

presented in Section 4. Section 5 shows the idea of
Business Process Management stands for a set of a component process repository, while in Section 6 the
techniques which aim to design, analyse, implement and main guidelines of sub-diagrams merge were described.
improve organisational processes. Process models can
help organisations in visualising processes of the
company and give opportunity for optimisation of the
process structure. Owing to this, the company may
achieve business goals in a more efficient way.
BPMN (Business Process Model and Notation) [1],
maintained by the OMG group, is the most widely
adopted notation for modelling business processes.
BPMN 2.0 is quite complex, as there are many
additional documents explaining the notation, such as
handbooks and papers devoted exclusively to the
pragmatics of the notation usage. In the paper, a brief
overview of the guidelines for process modelling as well
as various strategies of process design were provided.
According to the research based on industrial
applications [2], most of the manually created process
models are badly affected by quality issues. Therefore,
an automated support in the phase of process design is
highly appreciated. The proposed method may be seen as
a business process modelling technique based on
sub-diagrams generated as a result of a decomposition of
existing models. The workflow specification begins with
process elements and sub-processes of which the
Fig. 1. Overview of the decomposition method.
reusable sub-diagrams are created. These component
parts are then combined to form a comprehensive
process. An overview of the decomposition method is 2 Related Works
presented in Figure 1.
This paper is organised as follows. In Section 2, Decomposition of business process models is commonly
an overview of the existing solutions in the area of used in the literature to redesign processes or create new
process model design, as well as their applications, is models from existing components. One of the
presented. Section 3 describes in detail the idea of approaches consists in identifying maximal repeating
business process modelling and their mathematical fragments, called exact clones, which appear in multiple

*
Corresponding author: [email protected]
© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution
License 4.0 (http://creativecommons.org/licenses/by/4.0/).
ITM Web of Conferences 15, 01002 (2017) DOI: 10.1051/itmconf/20171501002
CMES’17

process models [3]. This is conducted by identifying The engineering process usually starts with business
SESE (Single Entry, Single Exit) structures in a process process modelling which constitutes a set of techniques
tree. Milani et al. [4] discuss criteria and heuristics for used to represent a certain workflow by providing visual
model decomposition. They also provide an evaluation diagrams depicting a business process with its
regarding the influence of these criteria on the dependencies. Thus, a business process model presents
understandability of selected models. The decomposition the way operations are performed to accomplish the
procedure may be supported by detecting names of non- planned objective of an organisation.
uniformly specified process element in order to enable An example field of application for business process
a faster creation of consistent complex models [5]. models is the specification of Manufacturing Execution
Another solution to the discussed problem was presented Systems (MES) [18]. Building a complete modelling
in [6] where business processes were presented as framework for the integration of a system technical and
a hierarchical ontology of business tasks apt for re- functional model, as well as production process model
composition. Such a description may simplify the ensures the understandability of the specification for all
dynamic creation of models of collaborative business the participants involved in the implementation process.
processes.
This paper is also related to structured and formal
3.1 Business Process Model and Notation
process representations which facilitate process
modelling, as well as their reconstruction. One of the Business process models in BPMN notation are
user-friendly approaches is the spreadsheet-based represented by diagrams composed of a limited set of
representation [7] where the set of activities is listed in graphical symbols [19]. Visualisation of activity flows
a table with specified branches and routing conditions. allows business users and developers to understand
Process models apt for re-composition can be also processes more clearly. It is a transparent visual tool
described in a declarative specification [8], which dedicated to modelling complex processes.
focuses on constraints and dependencies between BPMN model elements can be divided into four
activities instead of determining an explicit sequence major groups: flow objects, connecting objects, swim
flow. They can be also represented as a randomly- lanes and artefacts. Flow objects represent the principal
ordered list of tasks, along with their input and output elements of the process diagram. In this set, three major
conditions, which are then used for a constraint-based object types can be distinguished [1]:
process composition [9]. Another approach is a semantic - activities (rounded rectangles) which represent a work
model for BPMN specified in the CSP process algebra that must be performed in the process,
[10]. This is especially useful for specifying time - events (circles) which represent incidents occurring at
constraints between process activities. the time of process execution,
Process model repositories are an important tool - gateways (rhombi) which control the flow of tokens in
in the area of business process redesign. They can be the process.
developed in a form of a database storing relevant There are two kinds of activities: tasks and sub-
process fragments with a reference to the process processes. A task is an atomic action that can be
metadata [11]. This approach leads to an automated performed by a person or executed as an automated
generation of complex models. In case of such activity. On the other hand, a sub-process is a compound
a composition, the correctness of the generated process activity representing a complex work, divisible into
model must be ensured. Therefore, it is necessary to smaller parts which are defined as a separate process at
mention the identification and evaluation of issues the lower level.
regarding the flow consistency [12] as well as the layout The set of events covers three types of elements: start
of the generated business process model [13]. The events which indicate the beginning of a process with
automated process modelling should also include optional triggering conditions, end events that determine
verification whether the created model contains any the end of a path in a process or subprocess, as well as
typical anomalies such as deadlocks, livelocks or the intermediate events which may happen at any time
lack of synchronisation [14] [15] or event anomalies within a process.
[16]. The analysis of the formal aspects of a BPMN Gateways control the process flow by splitting it into
model can be also conducted using Alvis Modelling branches and joining them afterwards. The most typical
Language [17]. BPMN gateways are data-based, which means that they
evaluate an expression of process data, while they cannot
3 Business Process Models decide regarding the flow of the process [20]. Gateways
can be exclusive (XOR) which enable only one outgoing
Highly developed software systems are present in many sequence flow based on given process data, inclusive
areas of industrial and business applications, which (OR) that precede potentially parallel branches, as well
range from human-machine interface systems, through as parallel (AND) used to represent concurrent flows
production management, to web services covering executed without any conditions.
a large domain of implementations. To ensure the A set of the most typically used BPMN flow objects
reliability and efficiency of such systems and simplify is presented in Figure 2.
their design and development phases, advanced process
engineering approaches are in use.

2
ITM Web of Conferences 15, 01002 (2017) DOI: 10.1051/itmconf/20171501002
CMES’17

3.3 Mathematical Model of a BPMN diagram

To use the proposed process decomposition method, it is
necessary to transform a BPMN model into a form of
a graph. Therefore, a formal model of the process should
be determined before the conversion. A simple BPMN
model can be mathematically represented as a tuple
ℙ = (𝕆𝕆, 𝔽𝔽) [23] where:
- 𝕆𝕆 is the set of flow objects, such as: activities
(including task and sub-processes), events and gateways,
- 𝔽𝔽 ⊂ 𝕆𝕆 × 𝕆𝕆 is the set of sequence flows.
For the decomposition method, a task with boundary
events is represented in the formal model as a single
flow object having multiple outputs. In addition, the
proposed approach includes neither message flows nor
Fig. 2. Set of common BPMN flow objects. Based on [20].
artefacts, as the resulting sub-diagrams are generated
only for one determined process participant and
All the flow objects need to be connected using so- additional information may differ depending on
called connecting objects of two major types: a specific process model.
- sequence flow which shows the execution order of
a process, 3.4 Conversion to a Business Process Graph
- message flow which represents the exchange of
messages between process participants. Directed graphs are one of the common representation
To distinguish business functions or parts of the for business process models [24]. Since the proposed
system related to certain flow objects, the concept of decomposition approach considers only the layout of the
swimlanes is usually used in BPMN. Business process process model and not its execution, it was possible to
participants, which can either be entities within an generalise the existing model of atomic and complex
organisation or different collaborators in a process [21] process graphs. In the proposed structure called business
are represented by pools. Sub-partitions of a pool, which process graph, the existence of loops as well as multiple
represent specific objects or roles, are called lanes. They start and end events is allowed. Therefore, there are no
organise and categorise flow objects, connecting objects restrictions regarding the cyclicity of the graph and
and artefacts in a pool. Lanes are present in BPMN degrees of its vertices. In this case, the resulting graph
diagrams in a form of a rectangle extended either model may serve as a precise copy of a BPMN diagram.
vertically or horizontally along the length of a pool. Based on the formal model presented in Section 3.3,
Artefacts show additional pieces of information a business process graph can be defined as a connected,
which enable process designers to include more details directed graph GP = (VO , EF ) where:
in a model. Three main types of artefacts were defined in - 𝑉𝑉𝑂𝑂 is a non-empty set of vertices representing all flow
the standard [1]: data objects, groups and annotations. objects in a process,
- EF is a non-empty set of edges representing all
3.2 Running Example sequence flows in a process.
Transforming the BPMN model into a business
To illustrate the proposed method, a sample supply process graph consists in listing all process flow objects
process was selected. Its BPMN model is presented in along with the following properties:
Figure 3. The process goal is to purchase goods in case if - Object ID and Name (optional),
the available inventory is below the desired value. - Pool/Lane,
After the start, a warehouse employee checks the - Object Type (task, XOR split gateway, etc.),
inventory: if the available quantity is above the minimal - Condition (optional, used for OR and XOR gateways),
value, the process is ended; otherwise a requester creates - Input/Output Data Entities (optional).
a purchase order which is then reviewed by a supervisor. All the objects represented by vertices should be then
If the created order does not fulfil the requirements, it connected with directed edges which correspond to the
should be reprocessed by the requester, otherwise the sequence flow of the process. A directed edge 𝑒𝑒 ∈ 𝐸𝐸𝐹𝐹 is
funds are reserved by the accounting department and the a pair:
supervisor sends the order to the supplier. If the product
is not available at the supplier’s stock, a negative (𝑣𝑣1 , 𝑣𝑣2 ) ∈ 𝑉𝑉𝑂𝑂 (1)
response is sent to the supervisor and the purchase order
is cancelled. Otherwise, the customer receives an invoice where 𝑣𝑣1 represents its starting point and 𝑣𝑣2 its endpoint.
and a packing slip, which need to be recorded before In the business process graph edges contain also optional
releasing the funds and issuing the payment. As soon as information about a condition value, provided in case of
the payment is sent to the supplier, the process is alternative flows starting with XOR and OR gateways.
completed.

3
ITM Web of Conferences 15, 01002 (2017) DOI: 10.1051/itmconf/20171501002
CMES’17

Fig. 3. Sample model of a supply process. Based on [22].

In the next step, it is necessary to define adjacency

matrix D [25] of the generated business process graph. 4.2 Tri-gram Decomposition
Assuming that the analysed process consists of n flow
A more complex case occurs when there is a need to
objects, D is a square matrix of size n x n, defined by
generate sub-diagrams consisting of three process flow
Formula (2):
objects which can be connected to multiple elements. In
𝐷𝐷 = (𝑑𝑑𝑖𝑖𝑖𝑖 ) (2) such a situation, a component trigram for an object 𝑣𝑣 can
be in one of the following forms:
where rows and column indexes 𝑖𝑖, 𝑗𝑗 = 1. . 𝑛𝑛 correspond 1. 𝑣𝑣 and its two direct predecessors,
to the IDs of process objects. Each element of matrix 𝐷𝐷 2. 𝑣𝑣 with one predecessor and one successor,
is assigned to a value according to Formula (3): 3. 𝑣𝑣 and its two direct successors.
Providing the process graph 𝐺𝐺𝑃𝑃 = (𝑉𝑉𝑂𝑂 , 𝐸𝐸𝐹𝐹 ), having
1, 𝑖𝑖𝑖𝑖 (𝑣𝑣𝑖𝑖 , 𝑣𝑣𝑗𝑗 ) ∈ 𝐸𝐸𝐹𝐹 𝑛𝑛 vertices, as well as its incidence matrix 𝐷𝐷, the
∀ 𝑣𝑣𝑖𝑖 , 𝑣𝑣𝑗𝑗 ∈ 𝑉𝑉𝑂𝑂 ∶ 𝑑𝑑𝑖𝑖𝑖𝑖 = { (3)
0, 𝑜𝑜𝑜𝑜ℎ𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 decomposition algorithm consists in determining a set 𝑆𝑆3
by performing the following operation for each vertex 𝑣𝑣
Such a matrix stores all the information regarding
of graph 𝐺𝐺𝑃𝑃 :
flow connections between objects in the analysed
1. Get the sets of all direct predecessors and successors
process.
of 𝑣𝑣, denoted as 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃(𝑣𝑣) and 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆(𝑣𝑣), respectively.
2. Calculate cardinality of the determined sets, setting 𝑛𝑛𝑝𝑝
4 Model Decomposition as the number of predecessors and 𝑛𝑛𝑠𝑠 as the number of
successors.
The process model decomposition algorithm consists in 3. If 𝑛𝑛𝑝𝑝 > 0 and 𝑛𝑛𝑠𝑠 > 0 then add to set 𝑆𝑆3 all subgraphs
finding the set 𝑆𝑆𝑘𝑘 of all the subgraphs induced by induced by a set of vertices:
𝑘𝑘 vertices, where 2 ≤ 𝑘𝑘 < 𝑛𝑛. In the first step, it is
necessary to define appropriate subsets of 𝑘𝑘 vertices. {𝑣𝑣𝑝𝑝 , 𝑣𝑣, 𝑣𝑣𝑠𝑠 } (5)
Then, an adjacency matrix of each subgraph is created
by deleting all rows and columns from matrix 𝐷𝐷 which where 𝑣𝑣𝑝𝑝 ∈ 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃(𝑣𝑣) and 𝑣𝑣𝑠𝑠 ∈ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆(𝑣𝑣).
do not correspond to vertices for the selected subset. The 4. If 𝑛𝑛𝑝𝑝 ≥ 2 then add to 𝑆𝑆3 all subgraphs induced by
following subsections present methods for generating a set of vertices:
sub-diagrams having two and three elements, as well as
the concept of a general solution. {𝑣𝑣𝑝𝑝1 , 𝑣𝑣𝑝𝑝2 , 𝑣𝑣} (6)

where 𝑣𝑣𝑝𝑝1 , 𝑣𝑣𝑝𝑝2 ∈ 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃(𝑣𝑣) and 𝑣𝑣𝑝𝑝1 ≠ 𝑣𝑣𝑝𝑝2 .

4.1 Bi-gram Decomposition
5. If 𝑛𝑛𝑠𝑠 ≥ 2 then add to 𝑆𝑆3 all subgraphs induced by a set
In the most basic example, the process graph is of vertices:
decomposed into all the possible pairs of vertices
{𝑣𝑣, 𝑣𝑣𝑠𝑠1 , 𝑣𝑣𝑠𝑠2 } (7)
connected by an edge.
where 𝑣𝑣𝑠𝑠1 , 𝑣𝑣𝑠𝑠2 ∈ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆(𝑣𝑣) and 𝑣𝑣𝑠𝑠1 ≠ 𝑣𝑣𝑠𝑠2 .
𝑆𝑆2 = {({𝑣𝑣𝑖𝑖 , 𝑣𝑣𝑗𝑗 }, (𝑣𝑣𝑖𝑖 , 𝑣𝑣𝑗𝑗 )): 𝑑𝑑𝑖𝑖𝑖𝑖 = 1} (4)

4
ITM Web of Conferences 15, 01002 (2017) DOI: 10.1051/itmconf/20171501002
CMES’17

In the analysed example presented in Figure 3, 21 presented in Section 3.4 – vertices of each subgraph
flow objects were identified. After running the represent flow objects and non-zero values of matrix 𝐷𝐷
decomposition algorithm for tri-grams 29 different correspond to the sequence flow of the process.
subgraphs were generated. An example subgraph is Each of the sub-diagrams can have a different
shown in Figure 4. number of inputs and outputs, according to the flow
objects from which it is composed. In general, flow
objects can be divided into five main groups:
1. Sources: elements that can contain only output
sequence flow, such as start events.
2. Sinks: elements that can contain only outputs, such as
end events.
3. SESE (Single Entry Single Exit): tasks.
Fig. 4. Three-element subgraph generated as a result of process 4. SEME (Single Entry Multiple Exit): split gateways,
graph decomposition. tasks with boundary events, subprocesses.
5. MESE (Multiple Entry Single Exit): merge gateways.
Properties of the flow objects corresponding to
Regarding a best practice of BPMN process
vertices of the generated subgraph are listed in Table 1
modelling [2], implicit splits and joins should be avoided
while Table 2 presents edges included in the subgraph.
and replaced by gateways. Therefore, if the decomposed
Table 1. Properties of flow objects. model is constructed correctly, all its tasks are SESE
elements, except those containing boundary events, as
Object ID 2 3 4
described in Section 3.3. Sub-Diagrams that contain at
least one SEME and one MESE object represent the
MEME type (Multiple Entry Multiple Exit).
Pool/Lane Warehouse Operator
Based on the defined groups, it is possible to
determine the number of potential inputs and outputs in
Object Type XOR split end event task
each of the generated sub-diagrams, by counting the
Inventory number of ingoing and outgoing edges of each process
Object Name -- Create PO
Available flow object and comparing it to its group. This number is
Inventory > used to define functions 𝜎𝜎𝐼𝐼/𝑂𝑂 , described by Formula 8:
Condition -- --
min
0, 𝑖𝑖𝑖𝑖 𝑡𝑡ℎ𝑒𝑒𝑒𝑒𝑒𝑒 𝑎𝑎𝑎𝑎𝑎𝑎 𝑛𝑛𝑛𝑛 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
Table 2. Properties of sequence flows. 𝜎𝜎𝐼𝐼 = { 1, 𝑖𝑖𝑖𝑖 𝑡𝑡ℎ𝑒𝑒𝑒𝑒𝑒𝑒 𝑖𝑖𝑖𝑖 𝑜𝑜𝑜𝑜𝑜𝑜 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 (8)
2, 𝑜𝑜𝑜𝑜ℎ𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
Source Object 2 2
Respectively, function 𝜎𝜎𝑂𝑂 is defined for exits.
Table 3 presents to which group a sub-diagram belongs
Dest. Object 3 4
depending on values of functions 𝜎𝜎𝐼𝐼 and 𝜎𝜎𝑂𝑂 .
Figures 5-7 present three sub-diagrams of different
Condition Value yes no
groups generated based on the example BPMN model
presented in Figure 3:
1. SESE (Figure 5) – one available entry for “Reserve
4.3 K-gram Decomposition
Funds” task and one for “Send Order” task.
The general problem can be solved by determining for 2. SEME (Figure 6) – one available entry for split AND
a given number 𝑘𝑘 < 𝑛𝑛, all 𝑘𝑘-element subsets of 𝑉𝑉𝑂𝑂 and gateway and two available exits, one per each task.
generating connected subgraphs of 𝐺𝐺𝑃𝑃 induced by these 3. MEME (Figure 7) – one available entry for
subsets. A related solution was presented in [26], “Reprocess Order” task, multiple available entries for
however in this paper, a use of constraint programming join XOR gateway and multiple available exits for
technique is proposed. “Review Order” subprocess.
Following input data should be prepared to solve
Table 3. Sub-Diagram groups depending on functions 𝜎𝜎𝐼𝐼/𝑂𝑂 .
a constraint satisfaction problem:
- decision variable: 𝑘𝑘-element vector 𝑠𝑠𝑘𝑘 representing
a subset of process graph vertices, 𝝈𝝈 𝑶𝑶 0 1 2
𝝈𝝈 𝑰𝑰
- constraints: all elements of 𝑠𝑠𝑘𝑘 are different and
0 subprocess source source
connected,
- domain: adjacency matrix of graph 𝐺𝐺𝑃𝑃 .
1 sink SESE SEME

4.4 Resulting Sub-Diagrams 2 sink MESE MEME

The generated subgraphs can be retransformed into

BPMN models by executing the inversed method

5
ITM Web of Conferences 15, 01002 (2017) DOI: 10.1051/itmconf/20171501002
CMES’17

To ensure the correct repository creation without

omitting any relevant process sub-diagrams, a user-based
component validation was proposed. The operator who
performs the decomposition is presented a set of
potentially inconsistent component models. The selected
sub-diagrams need to satisfy at least one of the following
Fig. 5. A SESE sub-diagram. conditions:
1. Presence of a split gateway without any output branch.
2. Presence of a join gateway without any input branch.
3. More than two swimlanes considered.
In the next step, the component processes are
classified considering two criteria: the potential number
of inputs and outputs, as well as the diagram similarity.
The former is used for grouping models according to the
values of functions σI and σO defined in Section 4.4. The
latter is performed by applying the vector model
technique for business processes [27]. In this case, it is
necessary to define three input entities for the algorithm:
- document collection: two sub-diagram graphs
𝐺𝐺1 , 𝐺𝐺2 ∈ 𝑆𝑆𝑘𝑘 ,
Fig. 6. A SEME sub-diagram. - set of terms Θ: the union of all graph vertices with their
properties listed in Section 3.4,
- weights: for each term θ ∈ Θ a value of 1 is assigned if
the corresponding vertex is a part of the selected
subgraph and 0 otherwise.
Considering these two criteria, selecting the
appropriate sub-diagram consists either in choosing its
group, as shown in Table 3 or specific flow objects that
Fig. 7. A MEME sub-diagram. should belong to the component. In the second case, the
user is shown a set of similar sub-diagrams, out of which
a final selection is needed.
5 Towards a Sub-Diagram Repository
The decomposition algorithm presented in Section 4 6 Merge Guidelines Outline
generates all the possible sub-diagrams of 𝑘𝑘 elements.
This solution ensures that every possible configuration Composition of the BPMN model using the repository of
of process elements is ready to be reused in new process reusable parts should always be conducted with respect
models. However, some of the generated solutions may to business process modelling guidelines to avoid typical
potentially cause process inconsistencies, as they do not anomalies [28]. Incoherent business process models may
contain enough information about the process flow. be affected by syntactical anomalies which include
Let us analyse one of the tri-grams generated based incorrect usage of activities, gateways, connecting
on the example software testing process model which is objects or swimlanes. Another group are structural
presented in Figure 8. In this case, the resulting process anomalies, related to the improper dynamic behaviour of
fragment will not be fully reusable because of the AND the process, for example the presence of deadlocks and
gateway without any output element assigned. Taking infinite loops.
into consideration the fact, that the process flow Considering the correctness of the composed process
description contains information about the gateway type, model, it is recommended that its generation is supported by
as stated in Section 3.4, such a tri-gram configuration a graphical tool connected to the repository. The interface
prevents from executing only one activity after the “Sign should suggest recommended connection using the position-
Agreement” task which limits the component utility. At based classification technique [29]. An example of a forward
the same time, the number of branches after the AND model completion is presented in Figure 9.
gateway is not specified which makes the sub-diagram On the other hand, the system should also prevent the
underdetermined. user from creating links between flow objects which may
lead to structural anomalies. A potential deadlock is
shown in Figure 10.

7 Conclusions
The approach presented in this paper offers a set of
activities which support creation of reusable component
Fig. 8. Tri-gram generated as a result of process model business process models. The proposed method uses
decomposition.

6
ITM Web of Conferences 15, 01002 (2017) DOI: 10.1051/itmconf/20171501002
CMES’17

complete BPMN diagrams as input, based on which the 7. K. Kluza, P. Wiśniewski, Spreadsheet-based
component models are generated. business process modelling. In: 2016 Federated
Such a decomposition algorithm may be used to Conference on Computer Science and Information
create a component model repository which facilitates Systems p. 1355-1358 (IEEE, 2016)
process redesign as well as provides a possibility to 8. F. M. Maggi, A. J. Mooij, W. MP van der Aalst,
avoid common modelling anomalies. "User-guided discovery of declarative process
In the future work it is planned to include the models." In: 2011 IEEE Symposium on
message flow in the decomposition which will increase Computational Intelligence and Data Mining,
the support of the method for various process elements. p.192-199 (2011)
Another valuable extension would be to allow generation
9. P. Wiśniewski, K. Kluza, M. Ślażyński, A. Ligęza,
of sub-diagrams representing a specific set of activities,
Constraint-Based Composition of Business Process
based on their semantic description.
Models. In: BPAI Workshop held in International
Conference on Business Process Management
(accepted)
10. P. YH Wong, J. Gibbons, Science of Computer
Programming 76.8, p. 633-650 (2011)
11. M. Skouradaki, V. Andrikopoulos, F. Leymann,
Representative BPMN 2.0 process model generation
from recurring structures, In: IEEE International
Conference on Web Services, p. 468-475 (IEEE,
2016)
12. A. Burattin, et al., Software & Systems Modelling,
p. 1-22 (Springer, 2017)
13. V. Bernstein, P. Soffer, Identifying and quantifying
Fig. 9. BP modelling recommendation based on forward model
completion technique. Recommended links were marked with visual layout features of business process models,
green colour. In: International Conference on Enterprise,
Business-Process and Information Systems
Modelling, p. 200-213 (Springer, 2015)
14. A. Ligęza, T. Potempa, Artificial Intelligence for
Knowledge Management with BPMN and Rules,
In: IFIP International Workshop on Artificial
Intelligence for Knowledge Management, p.19-37
(Springer, 2012)
15. A. Suchenia et al., Selected Approaches Towards
Taxonomy of Business Process Anomalies. In:
Advances in Business ICT: New Ideas from Ongoing
Research. p. 65-85 (Springer, 2017)
16. A. Suchenia, A. Ligęza, International Journal of
Computer Technology & Applications 6.5, p. 789-
Fig. 10. Forbidden connections between sub-diagrams. 797 (2015)
17. M. Szpyrka, G. J. Nalepa, A. Ligeza, K. Kluza,
References Proposal of Formal Verification of Selected BPMN
Models with Alvis Modelling Language. In: IDC, p.
1. OMG, Business Process Model and Notation 249-255 (2011)
(BPMN): Version 2.0 specification. Technical
18. M. Ricken, B. Vogel-Heuser, Modelling of
Report (Object Management Group, 2011)
manufacturing execution systems: An
2. H. Leopold, J. Mendling, O. Günther, IEEE interdisciplinary challenge, 2010 IEEE Conference
Software 33.4, p. 26-33 (2016) on Emerging Technologies and Factory Automation
3. M. Dumas, L. García-Bañuelos, M. La Rosa, R. (ETFA), p.1-8 (IEEE, 2010)
Uba, Information Systems 38.4, p. 619-633 (2013) 19. A. Ligęza, Pomiary Automatyka Robotyka 15.12, p.
4. F. Milani et al., Business & Information Systems 210-212 (2011)
Engineering 58.1, p.7-17 (Springer, 2016) 20. B. Silver, B. Richard, BPMN method and style.
5. A. Koschmider, E. Blanchard, User assistance for Vol. 2. (Cody-Cassidy Press, 2009)
business process model decomposition, In: RCIS, 21. A. Lodhi, V. Küppen, G. Saake, Scientific Journal
p.445-454 (IEEE, 2007) of Riga Technical University. Computer
6. R. KL Ko, E. W. Lee, S. G. Lee., IEEE Transactions Sciences 43.1, p. 27-34 (2011)
on Services Computing 5.2, p. 246-259 (2012) 22. BPM Initiative Model Collection, http://bpnmai.org/

7
ITM Web of Conferences 15, 01002 (2017) DOI: 10.1051/itmconf/20171501002
CMES’17

23. A. Ligęza, Decision Making in Manufacturing and models, In: Seminal Contributions to Information
Services 5.1-2, p. 57-67 (2011) Systems Engineering, p. 405-419, (Springer, 2013)
24. I. M. Weber, Semantic Methods for Execution-level 28. A. Suchenia, P. Wiśniewski, A. Ligęza, Overview of
Business Process Modelling: Modelling Support Verification Tools for Business Process Models. In:
Through Process Verification and Service 2017 Federated Conference on Computer Science
Composition. Vol. 40 (Springer, 2009) and Information Systems, p. 303-310 (PTI, 2017)
25. K. Ruohonen, Graph theory, Graafiteoria lecture 29. K. Kluza, M. Baran, S. Bobek, G. J. Nalepa,
notes, (TUT, 2013) Overview of Recommendation Techniques in
26. P. Vasiliu, P. Tiberiu, “Mircea cel Batran” Naval Business Process Modelling, In: Knowledge
Academy Scientific Bulletin, 18.2, p. 419 (2015) Engineering and Software Engineering (Würzburg
University, 2013)
27. B. van Dongen, R. Dijkman, J. Mendling,
Measuring similarity between business process