Proceedings of the Fifth International Conference on Inventive Computation Technologies (ICICT-2020)

IEEE Xplore Part Number:CFP20F70-ART; ISBN:978-1-7281-4685-0

Robotic Process Automation Through Advance

Process Analysis Model
Devansh Hiren Timbadia Parin Jigishu Shah Sughosh Sudhanvan
Computer Engineering Department Computer Engineering Department Computer Engineering Department
NMIMS University NMIMS University NMIMS University
Mumbai, India Mumbai, India Mumbai, India
[email protected] [email protected] [email protected]
Supriya Agrawal
Faculty of Computer Engineering
Department, NMIMS University
Mumbai, India
[email protected]

Abstract—Robotics Process Automation is an advanced ability to manage, cost reduction, insights and analytics,
technology that builds an intelligent software robot that can non-invasive technology, increased compliance,
emulate human interactions with a business process. RPA is scalability, and flexibility. Another feature that RPA
an efficient automated method where software agents provides is recording manual tasks and then imitate the
interact through a graphical user interface in a human-like
whole process [6]. There are many organisations that
manner. RPA has various applications in most industries like
banking and finance, human resources, healthcare, etc. In provide a platform and tools to implement RPA, such as
this paper, we have proposed our RPA process analysis UiPath [7], BluePrism [8] or Automation Anywhere [9].
model and have compared it with a traditional model using The flow of the paper can be described as follows. Section
various comparison parameters like frequency of change, II compiles the literature survey conducted. Section III
degree of complexity, time is taken, screen usage and volume gives some examples of RPA being used in daily life.
of transactions and found out that proposed method gave Section IV defines process analysis and process model.
efficient results over the traditional method. Section V explains the proposed process model in the form
of a case study. Section VI summarises and compares the
Keywords—Robotic Process Automation (RPA), Business
Process Analysis, Software Agent, Process Model, Automation
existing process model and the proposed process model.
Model, Software Process Automation Section VII is the final conclusion of our study.
II. LITERATURE REVIEW
I. INTRODUCTION
In this study, we will be addressing the model
This Robotic process automation (RPA) is the practice described by Leshob, Bourgouin, and Renard [10] as
of automating routine business process with ‘software ‘traditional model’. This model decides whether the
robots’ that perform various tasks automatically. These process under consideration is to be automated or not by
tasks include transaction processing, IT management and considering three main factors, namely, eligibility,
automated online assistants. Robotic process automation is potential and relevance. The final decision is the
software robots that could effectively emulate human combination of the results of the above-mentioned factors.
beings by using artificial intelligence methods [1]. The For example, if a given process is eligible, with a
robotic process automation (RPA) represents neither potential=1 and relevance=0.3, then that process is
physical nor mechanical robot, even if it envisions some moderately suitable for automation.
electromechanical machine. In the term of robotic process A case study, analysing the application of RPA inside a
automation, robot refers to a software-based solution, BPO unit, was conducted by Aguirre and Rodriguez [11].
programmed to carry out procedures, processes or tasks on Their study yields that, when RPA was applied for some of
the repetitive way that are usually done by humans [2][3]. the organisation’s back-office tasks, an increase in
RPA can be applied in various functions like finance and productivity was observed with a reduced amount of
accounting (F&A), procurement, human resources, contact processing time.
centres, etc. Branches of RPA extends from sectors like A method that could be used for modelling of
banking to healthcare. In the banking and financial services processes is called the Integrate Definition (IDEF) group
industry [4], the RPA potential for F&A, card activation, of modelling methods. For business modelling, IDEF0
and fraudulent claims discovery is high whereas the RPA (Function Modelling) and IDEF3 (Process Description
potential for human resource (like payrolls, hiring Capture Method) are proven to be the best choice. The
candidates) is low. While in the healthcare industry [5], IDEF0 serves as a modelling language for analysis,
functions like report automation, system reconciliation has development and integration of processes and IDEF3
a higher RPA potential as compared to procurement (i.e. describes how a particular sub-processor a system works.
invoice processing, requisition-to-purchase). The benefits [12][13].
of RPA include rapid return on investment, enhanced
As stated by S. Al-Fedaghi [14], a process model is a
processes, better customer experience, elimination of method of representation of the reality of a process through
repetitive work, improved service delivery, enhanced

978-1-7281-4685-0/20/$31.00 ©2020 IEEE 953

Authorized licensed use limited to: UNIVERSIDAD POLITECNICA SALESIANA. Downloaded on July 16,2023 at 04:11:27 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Fifth International Conference on Inventive Computation Technologies (ICICT-2020)
IEEE Xplore Part Number:CFP20F70-ART; ISBN:978-1-7281-4685-0

diagrams that depict the flow of the process, its interaction downloading of a bank statement, finding any errors and
with databases and other entities. The most common way correcting them takes time, whereas a robot made by RPA
to represent a process is by designing a Business Process can do the same job in one-third of the time without the
Model and Notation (BPMN) diagram [15]. The major need of any human intervention. In this way, the
benefit of using such a standardised model is the employees can do more productive tasks in the time that
terminologies used, which can be easily applied to any and was spent on those repetitive processes. Also, the time
all kinds of business processes, of any complexity. Some spent on those processes can be greatly reduced by
other benefits include improved efficiency, agility and automating them.
clear insight.
A. Process Analysis
To better understand BPMN, we take an example
mentioned in the study done by Muehlen and Ho [16]. Process Analysis is a method to determine whether
First, the received complaint is analysed and then decided RPA can be applied to a particular process or not. It helps
whether to proceed for warranty verification or job in reaching to a conclusion for the most crucial decision
assignment. For warranty verification, it is directly sent to for a company, i.e. on which processes to invest money on
warranty claim, but for the job assignment, the required its automation. The selection of a process for RPA can be
type of service is selected. After the service is completed, a determined by various eligibility criteria like Potential,
quality assurance test is conducted and upon satisfaction, Complexity, FTE (Full Time Employ) savings, etc. Based
proceeded to collect the necessary funds. on these selected parameters and the arbitrary values
assigned with them, the eligibility is calculated. The main
III. REAL-WORLD APPLICATIONS role of process analysis is to choose a process in such a
RPA can be used on a day-to-day basis, some of the way that on automation, there is long term, fruitful
proposed processes are mentioned below: outcome bearing the ease of the company and increasing
the efficiency of the firm/company.
A. Procure-to-Pay B. Process Model
The procure-to-pay process involves a Purchase Order Process Model is a platform on which the
(PO), then the goods received and the final step is the implementation or execution of any process analysis can
payment step, involving creating invoices and logging all be done. A process model can only be formed when all the
the details to a database. Currently, this process is done by criteria of the process analysis are fulfilled. It provides a
systems like Enterprise Resource Planning (ERP) and meticulous and comprehensive diagnosis that acts as a
someone to enter all the necessary information. The
platform on which the process analysis for a process can be
manual intervention in this process can be automated using
executed.
an RPA robot on the front-end [17].

B. Automated Testing V. PURCHASE REQUISITION ANALYSIS: A CASE STUDY

Testing of new software is a very crucial but time- In this study, we consider a software robot (developed
consuming process. But, the most common test scenarios via an automation software) that will read an e-mail with a
can be easily automated using RPA and the process is pre-defined subject and download its attachments. Then, it
completed in no time, without any manual intervention reads the attached document via a pdf viewer and copies
[18]. all its data into a Microsoft Excel file and saves the file.
Then, it analyses each row and searches for that product on
C. Daily Profit & Loss Preparation the e-commerce website. This process is repeated for each
Many large trading companies need to track their P&L item in the attached document and a final excel sheet is
and risk exposures daily. Currently, the process is done prepared.
manually using MS Excel or other legacy tools. UiPath
The main flow of our proposed model can be
[19] claims that the handling time can be reduced by
summarised as shown in Fig. 2.
roughly 66% with 100% accuracy.

D. Process Mining
Process mining is a method used to analyse business
processes and extract relevant information from logs that
will help to detect undesired patterns, bottlenecks, etc.
With the help of automation, this process would be faster
and can be performed for large amounts of data,
continuously [20].

IV. WHAT ARE THE RPA PROCESS ANALYSIS AND

PROCESS MODEL?
There are numerous important processes that take place
in a company but repetitive maybe daily or weekly basis or Fig. 1. The basic flow of the model
a couple of days a week. These tasks can be automated
using the RPA approach. For example, the process of

978-1-7281-4685-0/20/$31.00 ©2020 IEEE 954

Authorized licensed use limited to: UNIVERSIDAD POLITECNICA SALESIANA. Downloaded on July 16,2023 at 04:11:27 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Fifth International Conference on Inventive Computation Technologies (ICICT-2020)
IEEE Xplore Part Number:CFP20F70-ART; ISBN:978-1-7281-4685-0

A. Automation Potential opened for each product request. So, here the automation
Automation potential is the capacity of the process that potential is Semi-Automation.
has the necessary abilities or qualities to become
successful or useful in the future. It is also a possibility of a
process to develop in a particular way. So, to determine the
type of automation for a process, the different tasks of the
process need to be analysed. Based on these criteria, some
parameters have been selected to determine the type of
automation to be implemented, as mentioned below.
1) Rules and Assumptions:
If, there are frequent changes expected in the system
in the next 6 months, or
the requirements can be handled through minor Fig. 3. Evaluating Automation Potential for the Process
changes in existing systems, or
the requirement be handled effectively by only B. Complexity
making a business process change, Complexity in simple terms is an assessment of the
tasks or events that make up the process and through which
Then, Automation Is Not Suggested
activity or a decision flow is decided before reaching the
Else if, the process requires complex cognitive output stage. Complexity characterises the behaviour of
capabilities, a system or model whose components interact in multiple
Then, High-Cost Automation Is Suggested ways and follow local rules, meaning there is no
reasonable higher instruction to define the various possible
Else if, Process involves manual decision making or interactions. Considering the type of interaction of the
process includes the use of physical-equipment/ automated process, the allocation of points is done as
security-token/ SMS-authentication, or mentioned below and then the level of complexity is
process involves unknown exceptions or determined, as shown in Fig. 6.
process involves uncertain triggers,
1) Rules and Assumptions:
Then, Semi-Automation Is Suggested For Each Application used, add 10 points to the
Else, Complete Automation Is Suggested complexity score.
The above-mentioned tasks are compiled into a Each Screen used in the process, add 2 points to the
generalized tabular form, as shown in Fig. 4, so that complexity score.
different processes can be easily categorized. For Each If-else condition used in the process, adds
2 points to the complexity score.
If, the net complexity score is less than 30,
Then, it is categorized as Low Complexity
If, the net complexity score is between 30 to 60,
Then, it is categorized as Medium Complexity
If, the net complexity score is greater than 60,
Then, it is categorized as High Complexity.

Fig. 2. Flowchart to Determine the Type of Automation Potential

2) Decision Making Process For Automation

Potential
This process involves manual decision making,
whether to purchase the product or not. It involves
unknown exceptions such as network errors, product out of
stock or purchase amount exceeds the budget. Also, it
involves uncertain triggers like a new browser window is

978-1-7281-4685-0/20/$31.00 ©2020 IEEE 955

Authorized licensed use limited to: UNIVERSIDAD POLITECNICA SALESIANA. Downloaded on July 16,2023 at 04:11:27 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Fifth International Conference on Inventive Computation Technologies (ICICT-2020)
IEEE Xplore Part Number:CFP20F70-ART; ISBN:978-1-7281-4685-0

Fig. 4. Flowchart to Determine the Type of Complexity

2) Decision Making Process For Complexity

This process is a manual and repetitive process that
requires about 4 Java/web-based applications. The
number of screens used is 16 per process with about
25 if-else scenarios each of standard input type. There
is a 31% chance of exception scenarios. It requires
image reading skills but no reasoning/cognitive skills
or remote access required. Based on the above data,
the complexity of this process is 30 or low.

Fig. 6. Flowchart to Determine the Type of FTE Savings

2) Decision Making Process For FTE Savings

The process involves 10,000 volumes of data per
process run, which takes about 1 hour of time. It is run on a
daily basis daily and engaging 2 employees in the process,
which sums up as about 720 hours yearly. Based on the
Fig. 5. Evaluating Complexity for the Process above data, the FTE savings for this process is 0.341 or
medium.
C. FTE Savings
FTE Savings stands for Full-Time Employee Savings.
It gives us information about the full-time employees hours
saved to execute the process. It also helps to budget,
economize and save money for the future. In the view of
time taken by a process or the amount of data it handles,
the FTE savings is calculated as shown in Fig. 8.
1) Rules and Assumptions:
Assuming that a single bot is used to automate at least
5 processes and the minimum FTE expected from a bot is
0.2.
If, automation saves less than 0.2 FTE,
Then, it is categorized as Low FTE Saving
If, automation saves between 0.2 to 0.6 FTE, Fig. 7. Calculating Full-Time Employment Savings for the Process

Then, it is categorized as Medium FTE Saving To put the process in a nutshell, we can say that the
If, automation saves more than 0.6 FTE, automation potential of the given process is Semi-
Automation, the complexity of the process has a collective
Then, it is categorized as High FTE Saving score of 30 due to which the process has Low complexity
and finally the FTE savings for the given process after
considering all the conditions are set to Medium. In the
end, the process, on the whole, falls in the Quick Win
Quadrant which acquaints us that the process is suitable for
Automation as shown in Fig. 9.

978-1-7281-4685-0/20/$31.00 ©2020 IEEE 956

Authorized licensed use limited to: UNIVERSIDAD POLITECNICA SALESIANA. Downloaded on July 16,2023 at 04:11:27 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Fifth International Conference on Inventive Computation Technologies (ICICT-2020)
IEEE Xplore Part Number:CFP20F70-ART; ISBN:978-1-7281-4685-0

dynamic, organisational view and information view, the

process is standard. Thus, the process is eligible for RPA.
Step II: The given process reduces human interaction
with the software as well as it follows the pre-defined
business rules. So, the process has potential = 1.
Step III: In this case study, the transaction volume is
high and the degree of complexity is moderate. So, the
process has relevance = 0.5.
Step IV: Considering the high value of relevance and
the potential of the process, it is highly suitable for
automation.

B. Proposed Model
The proposed model primarily takes three factors into
consideration.
 The first factor, Automation Potential evaluates and
Fig. 8. Summary Report tell the user what type of automation is suggested.
 Second, Complexity analyses criteria like cognitive
VI. RESULTS AND DISCUSSION skills required or exception possibility, etc and
To analyse the efficiency of the proposed model, we gives the complexity of the given process.
have compared it with the traditional model.  Third, FTE Savings takes into account the amount
of data, man-hours and the processing time required
A. Traditional Process Model to give the analysis from a business point of view.
According to [10], this model comprises of four-steps After analysing the parameters, some key parameters
to analyse a business process. The first step checks for the were taken into consideration and plotted according to
eligibility by considering the maturity and standardization their significance on the process model, shown in Fig. 11.
of the process. The next step analyses the potential of the
business process by the percentage of manual and software
interaction with respect to the whole process. The third
step assesses the relevance of automating a business
process with the RPA approach. The fourth and final step
classifies the process and provides the suitability of the
process for RPA, based on the result of the previous steps.
After a thorough examination of all the parameters,
some key parameters were selected and their weightage on
the entire process model is plotted in the form of a graph,
Fig. 10.

Fig. 10. Parameters' Weightage of Proposed Model

1) Assessment of Proposed Model:

Step I: The given process involves decision making,
unknown exceptions and uncertain triggers hence the
automation potential is semi-automation.
Step II: For the given process there are 4 Java/web-
based applications, 25 if-else conditions,16 screens used
per process and 31% chance of exceptions which
altogether makes complexity score low.
Step III: Volume of data for the given process is 10,000
which takes 1 hr of time to be processed with 2 employees
Fig. 9. Parameters' Weightage of Traditional Model
working daily which sums up to 720 hrs and based on this
data the FTE savings is medium.
1) Assessment of Traditional Model:
As per the method described in [10], we have, Step IV: Since the complexity of the given process is
low and FTE savings is high, the process falls in the quick
Step I: As the given process can be used for a long win quadrant
period of time and the software, it is mature. As the
process has some defined views, like functional view,

978-1-7281-4685-0/20/$31.00 ©2020 IEEE 957

Authorized licensed use limited to: UNIVERSIDAD POLITECNICA SALESIANA. Downloaded on July 16,2023 at 04:11:27 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Fifth International Conference on Inventive Computation Technologies (ICICT-2020)
IEEE Xplore Part Number:CFP20F70-ART; ISBN:978-1-7281-4685-0

C. Comparison of the traditional and proposed model decide if it is suitable for automation or not. The proposed
For better understanding the difference between model is comparatively better than the traditional model
traditional and proposed process models, their common because it deals with more number of parameters where
parameters are being taken into consideration and their each parameter has its own specifications. The proposed
weightage (i.e. the influence of a particular parameter over model also analyses the process from almost every
the whole process model) is as mentioned below. conceivable angle to give a detailed description with
improved accuracy whereas the analysis made by the
traditional model is imperfect and one-dimensional.
TABLE I. WEIGHTAGE COMPARISON OF COMMON PARAMETERS
BETWEEN THE TWO MODELS Although this work is at an early stage, we wish to
continue working on the same. Our next challenges are: (i)
Comparison Traditional Proposed improving the model by considering more parameters and
Parameters Model (%) Model (%) factors like cost estimation, paybacks, and benefits (ii)
Frequency of extend this model to support more business patterns (iii)
19 4.875 expand the empirical base of the process model.
Change
Degree of
9.5 9.75 REFERENCES
Complexity
Volume of
9.5 4.33 [1] Kampik, Timotheus, and Peter Hilton. "Towards Social Robotic
Transactions Process Automation." SIAS Conference 2019
Time Taken 9.5 8.667 [2] JovanoviÄ, Stefan Z., and S. Ä. Jelena. "Robotic Process
Automation: Overview And Opportunities." International Journal"
Screen Usage 9.5 3.9 Advanced Quality" 46.3-4 (2018): 34-39.
[3] van der Aalst, Wil MP, Martin Bichler, and Armin Heinzl.
Total 57 31.522 "Robotic process automation." (2018): 269-272.
[4] Romao, M., Costa, J., & Costa, C. J. (2019, June). Robotic Process
Automation: A Case Study in the Banking Industry. In 2019 14th
Iberian Conference on Information Systems and Technologies
(CISTI) (pp. 1-6). IEEE.
[5] Ratia, M., Myllärniemi, J., & Helander, N. (2018, October).
Robotic Process Automation-Creating Value by Digitalizing Work
in the Private Healthcare?. In Proceedings of the 22nd
International Academic Mindtrek Conference (pp. 222-227). ACM.
[6] Cewe C., Koch D., & Mertens R. (2017, September). Minimal
effort requirements engineering for robotic process automation
with test driven development and screen recording. In International
Conference on Business Process Management (pp. 642-648).
Springer, Cham.
[7] Tripathi, A. M. (2018). Learning Robotic Process Automation:
Create Software robots and automate business processes with the
leading RPA tool–UiPath. Packt Publishing Ltd.
[8] Chappell D. (2017). Introducing Blue Prism: Robotic Process
Automation for the Enterprise (2. utg.). San Francisco: David
Fig. 11. Weightage Comparison Chart of Common Parameters between Chappell and Associates.
the two models [9] Smys, S., & Ranganathan, G. (2019). Robot assisted sensing,
control and manufacture in automobile industry. J ISMAC, 1(03),
After comparison of the common parameters of both 180-187.
models, it is observed that the traditional model considers [10] Leshob, Abderrahmane, Audrey Bourgouin, and Laurent Renard.
19% weightage for the frequency of change, whereas for "Towards a process analysis approach to adopt robotic process
automation." 2018 IEEE 15th International Conference on e-
the proposed model its just 4.875%. Similarly, the Business Engineering (ICEBE). IEEE, 2018
weightage of the parameters, degree of complexity, [11] Aguirre, Santiago, and Alejandro Rodriguez. "Automation of a
transaction volume, time is taken and screen usage is 9.5% business process using robotic process automation (rpa): A case
each in the traditional model, but for the proposed model study." Workshop on Engineering Applications. Springer, Cham,
they are 9.75%, 4.33%, 8.667%, 3.9% respectively. 2017.
[12] Whitman, L., Huff, B., & Presley, A. (1997, January). Structured
models and dynamic systems analysis: the integration of the
VII. CONCLUSION IDEF0/IDEF3 modeling methods and discrete event simulation. In
Robotic Process Automation (RPA) is a targeted Winter Simulation Conference (pp. 518-524).
solution for optimizing human usage on mundane tasks, [13] Duraeva, O., Dmitrieva, E., & Tsapko, G. (2001, June). Activity
analysis of trade firm by IDEF0-methodology. In 5th Korea-Russia
unlike the unpopular notion of "workforce reduction". A International Symposium on Science and Technology. Proceedings.
process automation project needs process modelling as a KORUS 2001 (Cat. No. 01EX478) (Vol. 1, pp. 75-77). IEEE.
pre-requisite because it acts as a blueprint for the project. [14] Al-Fedaghi S. (2017). Business process modelling: blueprinting.
The Proposed Model provides a complete analysis of the arXiv preprint arXiv:1704.02973.
business process, which helps to calculate factors such as [15] Chinosi M., & Trombetta A. (2012). BPMN: An introduction to the
automation potential, complexity, and FTE savings of a standard. Computer Standards & Interfaces, 34(1), 124-134.
given process. Various parameters such as frequency of [16] zur Muehlen, M., & Ho, D. T. (2008, January). Service process
change, complex cognitive capabilities, etc. are taken into innovation: a case study of BPMN in practice. In Proceedings of
consideration. Using these factors and parameters, we the 41st Annual Hawaii International Conference on System
Sciences (HICSS 2008) (pp. 372-372). IEEE.

978-1-7281-4685-0/20/$31.00 ©2020 IEEE 958

Authorized licensed use limited to: UNIVERSIDAD POLITECNICA SALESIANA. Downloaded on July 16,2023 at 04:11:27 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the Fifth International Conference on Inventive Computation Technologies (ICICT-2020)
IEEE Xplore Part Number:CFP20F70-ART; ISBN:978-1-7281-4685-0

[17] Borthick, A. F. (2012). Designing continuous auditing for a highly

automated procure-to-pay process. Journal of Information Systems,
26(2), 153-166.
[18] Wu, X., & Sun, J. (2010, May). The study on an intelligent
general-purpose automated software testing suite. In 2010
International Conference on Intelligent Computation Technology
and Automation (Vol. 3, pp. 993-996). IEEE.
[19] Dumitrica, Gratiela. “Attended Automation in Action: Three Real
World Scenarios”. https://www.uipath.com/blog/three-use-cases-
for-attended-automation. (2017, November).
[20] Geyer-Klingeberg J., Nakladal J., Baldauf F., Veit F., van der Aalst
W. M. P., Casati F., Conforti R., de Leoni M. & Dumas M., (2018,
September). Process Mining and Robotic Process Automation: A
Perfect Match. In BPM (Dissertation/Demos/Industry) (pp. 124-
131).

978-1-7281-4685-0/20/$31.00 ©2020 IEEE 959

Authorized licensed use limited to: UNIVERSIDAD POLITECNICA SALESIANA. Downloaded on July 16,2023 at 04:11:27 UTC from IEEE Xplore. Restrictions apply.