Case Analytics Workbench: Hybrid Process Model Article Review Report

This report provides a comprehensive review of the research article titled 'Case Analytics Workbench: Platform for Hybrid Process Model Creation and Evolution,' which was presented at the 13th International Conference on Business Process Management. The article introduces a novel approach to business process modeling by creating a hybrid model that combines declarative and imperative processes. The review delves into the article's content, including the authors' discussion of the Case Analytics Workbench, an end-to-end system that uses imperative and declarative process mining, human interaction through a cloud environment, and event log clustering. The report analyzes the strengths and weaknesses of both imperative and declarative modeling, highlighting the need for a hybrid approach to meet the demands of knowledge-intensive processes. The review also examines the five key features of the Case Analytics Workbench, including its use of event log clustering and its integration with IBM Case Manager. Furthermore, the report discusses the workbench's architecture, comprising four modules: case model management, data management, process mining, and evidence management. The review then explores two practical case studies from the insurance and healthcare industries, demonstrating the workbench's capabilities in creating and refining business process models. The report concludes by addressing the issues highlighted by the authors, such as the human effort required and the need for empirical evaluation, while emphasizing the advantages of the cloud-based architecture and the hybrid modeling approach. This review aims to provide a detailed overview of the article's contribution to the field of Business Process Management.

BUSINESS PROCESS ENGINEERING
Business Process Management (BMP) Article Review
Case Analytics Workbench: Platform for Hybrid Process Model Creation and Evolution
By
Yu, X; Li, X; Liu, H; Mei, J; Mukhi, N; Ishakian, V; Xie, G; Lakshmanan, G.T. and Marin, M.
Student Name
Student ID
University
Professor
Date
Student Name – ID 1

The article under review in this essay is titled ‘Case Analytics Workbench: Platform for
Hybrid Process Model Creation and Evolution’. This article was presented by Yu, X; Li, X; Liu,
H; Mei, J; Mukhi, N; Ishakian, V; Xie, G; Lakshmanan, G.T. and Marin, M. during the 13th
International Conference on Business Process Management. The article discusses a new
approach of creating and developing an effective hybrid process model that combines both
declarative and imperative business modeling processes. The importance of business process
modeling in business process management cannot be overemphasized. For a business to thrive in
a competitive environment, it must have properly structured models that make it easier to
execute, analyze and improve various processes of the business. This review will be discussed by
describing the content and intension of the article, analyzing the discussion and conclusions
made by the authors, identifying issues highlighted by the authors about the model developed,
discussing the relevance of the model created in business process management and writing a
conclusion of the entire article.
This article discusses a new approach of creating and developing hybrid process models
called Case Analytics Workbench. According to the authors of the article, this is an end-to-end
system that accelerates the process of creating and developing hybrid process models through
combination of imperative and declarative process mining, human interaction through a cloud
environment and event log clustering. Most of business modeling processes have been either
imperative or declarative. Imperative modeling processes include XPDL (XML process
definition language), BPMN (business process model and notation and web services BPEL
(business process execution language). These processes apply imperative paradigm where only
allowed activity flows are captured and any unspecified flow is rejected implicitly thus limiting
many BPM applications (De Giacomo, et al., 2015). Declarative modeling processes include
SCIFF, Declare and DCR. These processes allow all activity flows as long as specified
constraints are not violated (Goedertier, et al., 2015). But as a result of changing business
environments, customer demands, economic factors and technologies, many businesses
nowadays have knowledge-intensive processes that cannot be efficiently met with either
declarative or imperative modeling process. The knowledge-intensive processes largely depend
on knowledge and expertise of workers who perform different interrelated knowledge-intensive
executive tasks that are centered on large volumes of data and information and need greater
flexibility. Both the imperative and declarative modeling process have their own strengths and
Student Name – ID 2

weaknesses using each of these processes individually is not enough to meet the current business
process modeling requirements (Caron & Vanthienen, 2016). To overcome this challenge,
authors of this article decided to develop an approach that combines both the declarative and
imperative business modeling processes. Hybrid models improve service quality of clients
offered by BPM companies, consultants or specialists (Parody, et al., 2016); (Wang, et al.,
2013). Therefore the purpose of this article is to examine a new approach of creating and
developing an effective hybrid process model called case analytics workbench, which combines
both declarative and imperative business modeling approaches. The hybrid modeling approach is
able to modeling all processes of the business.
Looking at it from human effort perspective, creating and refining a hybrid business
process model for knowledge-intensive processes is a demanding task. To begin with, the
modeler must visualize the process model so as to have a picture of the final product. This gets
more challenging as the complexity and structure of the business becomes more complex.
Second, the model should identify and follow an appropriate learning curve so as to learn the
latest hybrid modeling guidelines and languages. Third, the modeler must ensure that his hybrid
model is up-to-date by running various process mining approaches and explaining the results
comprehensively so as to improve it. Last but not least, it is important for the modeler to check
other analytics approaches so as to customize his model even better.
There are five main features that make case analytics workbench a better business
process modeling method. First, this method combines mining results obtained from both
imperative and declarative processes and extract evidences backed by data then uses an original
hybrid model to synthetize the evidences automatically. Second, it applies a hybrid modeling
approach that is acceptable in the industry by modeling declarative parts and imperative parts
using CMMN and BPMN respectively, and considering their extendibility and applicability.
Third, it uses event log clustering for customizing the business process model created. Fourth, it
provides better and state-of-the-art interfaces that improves user’s visualization and interactions
when creating and developing a business process model. Fifth, it is integrated with IMB Case
Manager (IBM) – a case management product (Yu, et al., 2015).
The case analytics workbench comprises of four main modules: case model management
module, data management module, process mining module, and evidence management module,
Student Name – ID 3

as shown in Figure 1 below. Case model management module is the most fundamental module
and it comprises of three components: case model storage, case model manager and case model
editor. This module provides users with a platform for interacting with a case model and allow
them to create, edit, save and transform business models (Yu, et al., 2015). Data management
module handles different process execution logs or records, convert them to approved event
record formats or layouts, and process them further with clustering. Process mining module uses
data management module’s event logs to run imperative and declarative process mining
methods. Evidence management module obtains mining results from process mining module and
presents it to the user for value-added understanding and insight. This model is also used to store,
visualize, filter and synthesize evidences.
In terms of organization, the workbench is divided into two segments. The first is the
server-side components that are set out in the cloud, and the second is the client-side components
that acts as an interface for users to execute their actions and also maintains business logic.
Figure 1: Advanced architecture of case analytics workbench (Yu, et al., 2015)
To demonstrate the capability of case analytics workbench on creating and developing a
business process model that has been created in this article, the authors performed two practical
case studies, one from insurance industry and another from healthcare industry. The first case
study was used to show how different components of the case analytics workbench are arranged
and synchronized to help the modeler create and develop an underwriting model more
conveniently. The second case study was about improving a complex hybrid model. In the first
Student Name – ID 4

case study, 4300 execution logs of underwriting process were collected from an insurance
company. Data presentation component of the workbench was used to transform the data to
formal event logs. Event log cluttering was used to generate two clusters: cluster 0 with 2038
auto insurance underwriting cases and cluster 1 with 2267 property insurance underwriting cases.
Declarative process mining method was used to generate visualization results in a Dendrogram.
The results were presented in bars with different colors, lengths, black line position (showing
support and goal correlation), as shown in Figure 2 below. The results provided very meaningful
and reasonable elements that were not well-defined in raw data, such as additional tasks and
constraints.
Figure 2: Visualization of results from declarative process-mining method (Yu, et al., 2015)
Imperative process-mining method is also used to mine sensible sub-processes for
various tasks then transformed them to evidences, as shown in Figure 3 below, which also
contains evidences transformed from results of declarative process-mining method. Before
evidence synthesis engine was automatically run, a filter was used to select strong evidences
with support and goal correlation greater than 0.3 and 0.5 respectively.
Student Name – ID 5

Figure 3: The main workbench user interface (Yu, et al., 2015)
Recommendations for model improvement were provided by evidence synthesizer
component, after which the user selected and confirmed the improvements in order to create the
final process model, which is shown in Figure 4 below. Therefore this case study demonstrated
data collection, preparation and clustering; process mining; and evidence analysis and synthesis
processes of case analytics workbench.
Student Name – ID 6

Figure 4: Final process model of the insurance underwriting process (Yu, et al., 2015)
The second case study was used to show care pathway improvement. In medical context,
a care pathway refers to a standardized process comprising of several care stages that correspond
to various conditions of disease progression, and each stage has its different clinical tasks and
unique constraints. Care pathways are usually complex and therefore the authors of this article
were optimistic that a hybrid process model would be an effective solution to the problem. In this
case study, the workbench was used to build a case model of initial care pathway derived from
clinical guide for managing cognitive heart failure. The authors used real electronic medical
records to mine evidences that were then used to refine the created model so as to generate one
that met the needs of a specific group of patients. The evidences obtained from the imperative
and declarative process-mining engines were meaningful and played a key role in refining the
care pathway model, which is shown in Figure 5 below.
Student Name – ID 7

Figure 5: Final process model of care pathway improvement (Yu, et al., 2015)
The case analytics workbench demonstrated in this article by its authors is evidently an
efficient, flexible and reliable method of creating and developing a business process model.
Nevertheless, the authors have highlighted some issues that must be considered to make the
process successful or even better. One of these issues is the human effort needed when creating a
business process model using the workbench. This process is knowledge intensive and requires
the modeler to visualize the entire process and final product at the start, learn the latest hybrid
modeling guidelines and languages any time they are released, continuously update the model
being created and find out what other modelers are creating so as to improve and customize his
model. Another issue is the cost of decoupling. If the volume of data being used is too large, it
becomes relatively burdensome for the network to transmit the data between different modules
of the model. Therefore advanced methods of data transmission are necessary. The authors have
also highlighted the need for empirical evaluation aimed at enabling different modelers create a
hybrid process model integrally. This will enable the methods developed to run models that have
already been generated instead of just being used as modeling tools.
The case studies presented in this article have demonstrated the capability of case
analytics workbench to create a hybrid business process model. This approach is cloud-based and
Student Name – ID 8

has coherent interface that enables users to interact with case models and check analysis results
absolutely hassle-free. This cloud-based architecture has numerous advantages including
extendibility and versatility. In this article, it has been proven to be very useful and efficient for
use in healthcare and insurance industries. The fact that hybrid process model combines
declarative and imperative process-mining approaches makes it better than discrete models such
as BPMN model. The hybrid process model created in this article is applicable in a wide range of
fields of BPM and it has been accepted by BPM product developers and managers, BPM
specialists, business process administrators, clinical physicians, etc. The key advantages
highlighted by its users include: combination of imperative and declarative process-mining
methods; involvement of clustering technique; and improvement of user interaction. Still there is
room for improving the hybrid process modelling and more focus should be put on empirical
evaluation so that different modelers can create a hybrid model integrally and use it to run
already generated models.
The authors of this article have comprehensively presented a solution of creating a cloud-
based, end-to-end system that combines imperative and declarative process mining to create a
hybrid business process model. Case analytics workbench, as it has been called, is an advanced,
flexible, user friendly and versatile hybrid business process model that is can be used to solve a
variety of business process management problems. The authors have successfully demonstrated
application of this model in healthcare and insurance industries. Even though this hybrid model
has numerous advantages over other models that use either imperative or declarative process-
mining methods, there is still great room for improvement. Some of the suggested researches
include: further empirical evaluation aimed at enabling different modelers to create a hybrid
process model integrally; improvement of data transmission mechanisms; and reduction of
decoupling cost. All in all, results obtained using the case analytic workbench developed and
presented by authors of this article were impressive and accepted by various specialists,
including BPM specialists, business process administrators, BPM product developers and
managers and clinical physicians. Therefore hybrid business process models such as case
analytics benchmark have great potential of improving business process management.
References
Student Name – ID 9

Caron, F. & Vanthienen, J., 2016. Exploring business process modelling paradigms and design-
time to run-time transitions. Journal of Enterprise Information Systems, 10(7), pp. 790-813.
De Giacomo, G., Dumas, M., Maggi, F. & Montali, M., 2015. Declarative Process Modeling in
BPMN. In: Z. J, K. M & J. P, eds. Advanced Information Systems Engineering . Cham,
Switzerland: Springer, pp. 84-100.
Goedertier, S., Vanthienen, J. & F, C., 2015. Declarative business process modelling: principles
and modelling languages. Journal of Enterprise Information Systems, 9(2), pp. 161-185.
Parody, L., Gomez-Lopez, M. & Gasca, R., 2016. Hybrid business process modeling for the
optimization of outcome data. Journal of Information and Software Technology, 70(C), pp. 140-
154.
Wang, Y., Huang, L. & Guo, Y., 2013. Intgrating declarative and imperative approach to model
logistics service processes. Journal of Industrial Engineering and Management, 6(1), pp. 237-
248.
Yu, X. et al., 2015. Case Analytics Workbench: Platform for Hybrid Process Model Creation
and Evolution. Innsbruck, Austria, 13th International Conference on Business Process
Management.
Student Name – ID 10