What email address or phone number would you like to use to sign in to Docs.com?
If you already have an account that you use with Office or other Microsoft services, enter it here.
Or sign in with:
Signing in allows you to download and like content, and it provides the authors analytical data about your interactions with their content.
Embed code for: A short theory on BPM
Select a size
Theories of BPM – An Analytical Approach
This project has not been sponsored by anybody or any company – I have been working on this for the last 5 years – finally I had the time to put something in a piece of paper.
Over last 5 to 6 years working with BPM (Business Process management) and BPA (Business Process Architecture) – Petri’s works have appeared to be most cohesive and logically sustainable – there has been some great books and theories around EA (Enterprise Engineering) specially in Netherlands and Belgium but there has not been much of a corollary which can extend Petri’s work in EA world.
Petri’s work has been cohesive and rational by nature because of the fact that it is scalable by nature – now with large scale transaction systems petri-net becomes more relevant. It provides the analytical methods necessary to model architecture of high volume transaction systems.
In order to enable an enterprise architecture, we need a method which can enable processes based on transaction pattern. Dr. Dietz’s work is firmly around the notion of transaction and the patterns of transaction.
Keywords: Information Triangle, Enterprise DNA, Meaning Triangle, Petri-nets, cellular-automata, block-chain,
Why do BPM – Business Process Management Work – please note the question is not “how” – but its “Why” – so we are not focusing on the mechanics but on the theory of why it works. The “why” defines the physics while “how” defines the mechanics or “chemistry”.
There is a paradox in industry today where some companies thrive on a set of so called static processes which are drawn on pieces of paper without the people part being sufficiently elucidated. The same set of processes which applied to other companies does not show any success and mostly miserable failures. This brings to 2 unknowns which are people and technology (the other 2 columns of the classical BPM theories). This also brings into question a 3rd unknown which questions the validity of these processes in a dynamic world. The processes change at the speed of technology and with new ideas like “Dev-ops”, scalable and rentable hardware technology can change on a dime. (some companies like Uber, AIRBNB are growing by leaps and bounds and introducing new processes based on customer demand so the concept of static processes cannot be that valid). So the next obvious question will be does the nature of the transactions and human behavior help design the processes or the other way around? For the sake of argument, we will consider that all compliances are built into the transactions. In other terms is nature of the transaction the leading or the lagging indicator.
A BPM Triangle
Business DNA – in order to define business characteristics, we need to define the business DNA which calls for a repetitive construct – business process architecture as done today is hardly as repetitive construct.
How Business DNA differs from Business Process Architecture – BPA is a very high level construct – more at the level where a human being interacts with its environment – it hardly defines the inner working of the business or how it develops and maintains its characteristics.
Transactions/Petri-nets and Business DNA – Petri-net models can act more like the skeletal level architecture inside the body and does not really represent the DNA at the cellar level. The cellular level architecture is yet to be defined.
How does business process look like at cellular level – in order to work for business DNA the cellular level architecture needs to be defined precisely.
Where does security and compliance sit here – like human antibodies (which mostly work with the blood spectrum) security and compliance sits at the cellar level and can communicate in between the cells – both of them has to be fluid – which means once a security has been set up for one role at one node – it will perpetuate for that role at all nodes.
Threat vs. opportunity – Risk Characterization happens at cellular level too – which means a threat needs to be identified and mitigated using a method at an atomic level during each transaction.
Hexagonal cell coverage of a business space – the business space is really a n-dimensional space but if for the sake of argument, we consider it to be a 3 dimensional space of people, process and technology then there must be an effective method to model the business in that continuum.
Tie to view points (with each dimension) – each person with a particular role that has a different viewpoint – they look at processes completely differently so this single process model
Challenges with existing Business Process Architecture
In today’s world the business process architecture developed have very little connection to business process management from a theory perspective – yes we do model to execute but those are very crude method of converting a model in a executable workflow – it has very little to do with the theories of BPM or canonical data or state of the data.
Business process architecture or Business Process Management tools and methods are not conducive to complexity (relationships with more than one variable) and adoption of process changes has very little organized method – it’s almost completely manual.
The fact of creating a computer or AI generated process based on a set of characteristics, roles and business behavior and business data is almost absent.
Regeneration of models using the same set of objects using changes to business characteristics (delta change)
If we look at the enterprise as a 3-D continuum, then we need to talk about area and space and not just some box and lines.
Each person working in an enterprise has a little n-dimensional multi-hedron around him – each node designates a role he or she may have – this is like a daisy ball of an IBM type writer –
With the current business process architecture there is no way to measure the external or internal forces (influences) on the processes – these are the things which characterizes the processes. There needs to be a method which not w=only can measure the forces on the processes
The triangle can be looked through various lenses – based on pure trigonometry (this is a static method) or through the forces (defining the sides of the triangle as in dynamics),
From Semiotics we get the meaning triangle
Figure 1 The Meaning Triangle
If you have to represent BPM using the information triangle or the meaning triangle of semiotics, then how will it look? Here is a basic form - may not be perfect but this is something to start with – this is just a linear mapping
The Meaning Triangle and the BPM triangle
Role – the role in a BPM triangle is the one “responsible” for executing the “method”
Data – The “data” in the BPM triangle is in ACID form which means Atomicity, Consistency, Isolation, Durability.
Method – Method can be construed as an atomic level “capability” - the method can be broken into an infinitely small element of an act (a verb).
Figure 2 Modified meaning triangle (BPM triangle)
This somewhat translates into
“Role” is a “concept” inside the boundaries of enterprise – that we human being create for our convenience – it is subjective by nature.
This indicates the “people” portion of a classical business model (people, process and technology)
“Data” is the “object” that roles work on – this is probably the most objective element by nature – a data can only be interpreted in one way in its acid form.
This designates the “process” section of a classical business model – this can be directly inferred from Petri-s models since data flows down the process architecture. (people, process and technology)
“Method” is a “sign” which is subjective at the same time open to some representation which can be interpreted.
The method resides inside technology or is being triggered with a set of technologies – which can be IT related, mechanical or manual. But it is a set of defined technology. (people, process and technology)
Now, if we go back to the earlier question of “why” – there can be various answers to this based on the type of “role” – as any operator or agent performs his/her/their duties within enterprise and interacts with the outside world for the enterprise they have acquire “Facts” and based on the “Facts” they or their assigned agent perform transactions which enables “Production”. The method ties to this “why” question. Any agent uses a selected “method” out of a few hundred or even thousands of method available to him – these “methods” at the atomic level create the foundation of the “capabilities”. Example: in a sales order if there are 10 line items then all the costs can be added and one line can be created with a total cost – the method used here is a “summation” of sales order line values but that is a “capability” at atomic level. (the definition of capability is beyond vague in industry and has very little mathematical foundation)
The important thing to understand here is that the triangle operates at an atomic level.
In the “Vector” form the triangle would look something like this - From the mathematics of it all – this vector form is probably most important while developing a computer model –
Figure 3 BPM Triangle Vector Form
Any person can have many “hats” inside an enterprise defining many “roles” – so a point definition of a person inside an enterprise can be something like this as shown below
Figure 4 Multi-dimensional role model
The model above also designates the Domains Plains for Business Process triangle -
Which means each individual inside the enterprise operates in multiple domains with multiple dimensions of responsibilities sometimes which have several domains combined. Each domain can be represented by a triangle as shown above. The triangles from multiple domains can be combined to create composite responsibilities.
A person “P” has access to 6 data from D1 to D6 – these accesses are not same – they are based on the methods applied in each transaction – which means in come case the person P may have only read access, while in others it can be write, update and in a few cases delete.
So in this table we can show that the person P using the various roles and data
Figure 5 A person with various roles accessing various data
The security model of this person in the IT system looks somewhat like this – this shows what type of access a person needs to what type of data and under what kind of role.
Figure 6 Composite role model for an enterprise employee
So in our BPM triangle model this can be modified into Access defining the security model and Utilization defining the trust model. (One of the important unanswered question here is that does the access to the method define the access to the data or the other way round).
Figure 7 Security & Rules Model
Now if we consider a particular data D1 then an instance of that data can be represented like this- which not only represents the life-cycle of the data but also introduces a new variable which is time. This helps create the state model.
Figure 8 Activities of roles in a phased manner
Using this Triangulation structure the business process space may look something like this (This is also similar to hexagonal models used in cellular world)
Figure 9 BPM Triangle Network
This is a graph in 3 -dimensional space and the following scales can be used – but these are absolute scales using people, process and technology – this does not have any “state” – so this model can be looked at as an absolute model – the bigger challenge will be assigning the scale for the 3 dimensions (people, process and technology).
Figure 10 People Process and Technology
The above figure in the 3-D space can be projected in a 2 dimensional space – the data entity can be followed in this graph through the transactions – so the transaction will be the leading indicator and the process map will be the lagging indicator in real time. But once a pattern has been recognized then the pattern will be the leading indicator for the
Figure 11 2-D BPM network
When we have to add “state” then typically we will use data and time as the 2 axis. The state of the data changes over time. This is more of less a binary function which means when the data is active and other times when it is in active. Some data (as its atomic level) are more active than other data (as for example the social security number of a person inside the government environment may be more active than their physical address). Each data set has its own chart and thus 2 charts can be combined and compared.
When we add the time factor to this chart and track that data entity then the graph looks like this
Figure 12 Combining Time phased model with network model
Which can further provide “a” dynamic process model – which has a state and changes with the passage of time.
Figure 13 A process model to show the data flow through the transactions
The business DNA structure will be something like this
Figure 14 A Basic Business DNA structure
The DNA sequence will look something more like this below – here in this case one transaction is using 1 single method to associate 2 data entities with 2 roles. (example of adding the item and the other is adding the price in a sales order)
Figure 15 A compound structure
The Business genetic sequence or footprint can be extended through either the method or data or role. So the model shown above can be extended in multiple ways and this creates the distinct genetic patterns of each company which help them to survive and thrive (or fail). Within this structure the adoptive characteristics of the business originates.
The “end to end” (like Hire to Fire) processes have a life cycle somewhat like this
Figure 16 A Business DNA model
This is something which the consulting companies try to sell as their knowledge base and industry standard or Best Known Method in the form of hundreds of process models but there is an inherent weakness to those static models which are
They do not define the character of a company – there is an elastic nature in that which means
They are not flexible of scalable.
They can be considered to be pseudo flexible when one draws another static process model to replace or augment a previous one but still the characteristic portion is absent.
The cross section of an enterprise landscape can be designated in to a 2 dimensional model (the hexagonal structure has been used for most efficient use of space)
Figure 17 Business Landscape
From this model the following results can be deducted - Collaboration, Association, Difference, Segregation
Figure 18 Business Relationships
Under the current business modeling techniques, we can easily define relationships in a tree type hierarchy but the current trend in industry is to have a flatter organization where most of the collaboration is knowledge based or need based. Above of all much of the knowledge based relationship within the organization happens organically. But theoretically it will be “informa” based roles interacting with “performa” based roles etc. This is also where some of organization theories and practices fall apart. (this is also true in case of matrix organization because within the same matrix there are typically the folks from the same organization level though different verticals).
Even though business silos can be drawn as end to end processes – in real world it rarely happens that way – rather small or even large groups work within small boundaries with some target deliverable – within those boundaries there can be 2 types of transfer between 2 roles which are either data or a method. The transfer of data is more around “forma” or shared transactions (example one person putting in the price and another adding the tax percentage etc. for the same object) but there can be collaboration on the method too which is more of a knowledge transfer.
Most of the process designs in a linear fashion was designed in a world of large legacy applications (like SAP R2 etc.) when systems had very little flexibility. So it was the job of the vendors to prescribe how to follow processes to make their tools viable and useful. Real life is far more flexible which rarely follows such in-flexible processes. Real life is mostly full of exceptions and in many cases because we cannot model all the variables. Few things are changing how we understand business process management:
Cheap Scalable transaction systems where actors and transactions can be designed and deployed in minutes rather than years (though this create a security nightmare)
The genome sequencing work by Craig Venter has changed the way we look computing for industrial purposes - this happened because of the increase in computing power. There will be a large change in process architecture like genome sequencing - we have to take advantage of computer processing power and deep learning from transactions. (There is a lot of history and characteristics embedded in the transaction history.)
The recent works in block-chain architecture where transactions are recorded in public registries change our outlook towards transparency, security as well as storage methods (there are active transactions and passive transactions and block chains can hold both types – in the legacy models like SAP all transactions are active)
The lack of process generation from transaction data is a symptom that we are using transactions only as lagging indicator and not as leading indicator to design processes.
Example: We consider a process as shown below
Figure 19 Business Process Model
In an ideal world the data would look like as shown below – there are 2 actors here the initiator (requestor) and the Approver (Reviewer).
Figure 20 A basic transaction using BPM triangle with 2 actors
There may be multiple data in one request and in the data table for this single transaction as shown in the table below – each of the data will have its own data table and may have sub-methods for validation (like address validation M11). The general method used here by Role R1 is M1 to produce the data D1. The transaction itself creates a record. This is the product of the triangle. This record can be stored in a system or can be produced as a document which can be transmitted to another system or data. But this particular triangle will not be changed or altered.
Table 1 Data record 1 from requestor
Once the data has been entered it becomes a permanent record in the transaction system but based on the past records we can fairly easily predict who will request for admission. The data can be seen through the BPM triangle in the following fashion (Figure 21)
Figure 21 Table 1 data in BPM triangle format
Based on the above diagram the basic transaction can be represented something as shown below (Figure 22) with 2 roles R1 and R2 interacting through multiple data in a single transaction. This can also be represented by 2 simple 3x4 matrix in an additive form. The matrix form is important because the data sets need to match up on both sides for the transaction to work (which in BPM world is often called the canonical model). We could have used the same data and method axis for the model below instead of using 2 separate axis but then the model would have become much more complicated.
Figure 22 The data model for a particular transaction using BPM triangle
The second transaction table looks something like below (Table 1)
Table 2 Transaction Table for receiver or approver
In real life a process would look something like this below where we will have 3 roles that of a requestor, reviewer and Approver – the second 2 roles can be played by the same person or actor.
Figure 23A standard transaction with 3 roles – each of the triangle is segregated from the others (SOD rule)
M1= Write / Update Data (Action Enter Data)
M2 = Read Request Data (action ensure correctness)
M3 = Read Request Data (action for Approval or disapproval)
M3 is a typically a binary function which means you accept or reject the request. M3 may have a 3rd “wait” state – where by the approver requests for more information. Segregation of each transaction is based on the fact that other roles cannot participate in these transactions. This leads to the security model with segregation of duties between roles.
Figure 24 Reverse direction transaction - this doesn't undo the existing transaction though
M4= Request Update from R1 by R3
M5 = Approve Request of R3 by R2
M6 = R1 Receive Update Request from R2
Step 3 = Step 1 (exactly same set of transaction)
After Step 1 or 3 there is a transformation where some more data is added to this existing data set D1. The resulting triangles look like this (Figure 1). This does not imply that data set D1 disappears. Since it is the primary source of truth rather a new data set D2 is created where a new set of data is augmented to D1. This is the product of a transaction with method M3. (Figure 25, Figure 26, Figure 27)
Figure 25 Completion of transaction
And the transformation looks like as below
Figure 26 The matrix addition (3x4 matrix)
The data set looks as below
Figure 27 Dataset D2
If we now take the entire process as shown below (Figure 28)
Figure 28 Complete Transaction set for the request and approval (or disapproval)
This is the simplest modular process which can be repeated again and again – most other processes can copy this part or build on it. The security patterns build on this process with a set of transaction needs to be modified heuristically to match a particular bigger process. Then we can get 2 kinds of basic BPM triangle model representing part of the DNA of the enterprise.
Figure 29 Basic DNA Structure of a company which can be used by all groups using any approval process
Figure 30 Basic DNA Structure of a company which can be used by all groups using any approval process (version 2)
The models in Figure 29 & Figure 30 can be represented in other arrangements but companies which are successful can choose the most efficient arrangement for them and replicate it. This does not mean that this process is most efficient universally rather it is best suited based on interactions with other processes and the human organization inside the company.
IEEE 1471/ISO 42010
Dr. Dietz – Enterprise Ontology
Transaction Processing – Bernstein
Mathematica – GraphData, Circlepoint,
A short Theory of BPM24
Running head: A short Theory of BPM1but based on the past records we can fairly easily predict who will request for admission. The data can be seen through the BPM triangle in the following fashion (Figure 21)
M3 is a typically a binary function which means you accept or reject the request. M3 may have a 3rd “wait” state –