Semantic BPM part two

In part one, we have looked at the limitations of current enterprise systems to build a social process ecosystem, in order to  architects, managers and company employees can understand what is the business environment, what means the concepts used within the organization and what is the business domain (context) all about.

I also created the quest of establishing a dynamic method for structuring business ontologies, taking into account that usually there is a misalignment between the enterprise architecture and execution of business processes, especially in an era when more and more processes are no longer structured as it used to be (described in flows, in documents, procedures).

In this part we define a method for discovery of processes semantics and how it contributes to the enterprise ontological management.

Unstrutured work increases the importance of semantics

The dawn of “Case Management” amplified even more the importance of building enterprise ontologies.

There are several synonyms for “Case Management” [39] like:

  • Unstructured work;
  • AD-HOC process;
  • Dynamic Case Management;

I prefer to use the term AD-HOC process because is much more comprehensive, the name “Case Management” seems to apply only to managing a dossier with files inside it, as happens for example in law firms, as well the terminology used seems to be very restrictive. The term AD-HOC is more related to the need to manage creative thinking and empowerment and I will stick to this definition.

Executing AD-HOC process may be supported in portals or wikis. This is not anything new because the technology has been available for long (you can do it with e-mail). For example using the enterprise portal and especially in process mashups [2], collaborating like in the example provided before using Google Wave.

While acknowledging that socialization has brought advantages in how people can collaborate to execute some task work being able to work together in collaborative environment, still lacks a method for ultimately people “understand” the artifacts they are dealing every day. This is particularly important because even despite for managers, process experts, there is a common dictionary of business concepts, for the other people, often these concepts remain abstract affecting the way work is performed.

Some years ago if we need to identify the relationships between business processes, we had to rely on graphical representation such as prescribed in SAP Modeling Handbook [11], but with the advent of AD-HOC process the classical representations are compromised since AD-HOC processes are not even mapped because they are executed seeking to achieve a particular outcome. The dynamics of the business world is increasingly running around unpredictable work requires a dynamic model of ontology representation, because employees manage informational entities (artifacts or objects) that have to be clearly defined, otherwise they are not sure the they are manipulating.

People are often confused with the concepts of output and result of a process, some people think this means the same but they not. To clarify what it is it’s worth revisiting the definition given by Arthur R. Tenner and Irving J. DeToro [12] back in 1996!

Output – Ability of the process comply with all applicable specifications.
Result – Ability of the outputs meet customer’s needs.

To understand the changing paradox processes can be executed, let us consider a process to select someone to fill a job vacancy within the Financial Division. For such a position there is no standardized process because it is not appropriate to the type of position he want to hire.

The person carrying the responsibility to do it must:

  • Define activities required to reach the desired result – hire someone.
  • Identify the informational entities (the artifacts, objects, or data if you want) that are manipulated during the execution of the activities, for example, Candidate, Interview, Recommendation, Qualifications, etc.

In the figure below we can see how a BPM tool, can let the user define what we wants to accomplish without any kind of flowchart.

AD-HOC HR Process

Each activity must be associated with the informational entities that will be manipulated at runtime.
However, when the process is already being executed is identified the need to add one more activity , in this case a review from a Human Resources Department. For a BPM tool with advanced capabilities that is not a restriction.

Inserting a new task dynamically

Today there is no single way to perform the work. Human judgment, unpredictable conditions, business rules determine which activities should be performed, designed in real time.

Peter Drucker [37] stated that working in the knowledge era means:

The first questions to increase productivity and work smarter, have to be: What is the task? What we are trying to achieve? Why do we have to do this anyway?

The model

The model for semantic business process management is presented below built using a UML class diagram because is the ultimate abstraction and it applies to every industry sector and process type. This model could not have been stutured without the contribution of André Brandão and João Graça [13].

Semantic BPM model

This model is based on the principles described in algebraic graph theory.

A network is based on the following principles:

It is a set of nodes and arcs. In this case the nodes are represented by processes and arcs representing the relationships between processes.

Process network
In this example there are four processes. Processes P1 and P2 and P1 and P3 are connected, P2 is linked with P3, P4 is connected to the process P3.

Each arc expresses a relationship. In this particular example of a process network as opposed to a social network there is only one type of relationship, which can be called “connected to”. In personal social network you can assign different relationships types among people such as: “work with”, “family” and I believe there is a temptation you can assign different names to the relations between processes such as: “support”; ” supported “,” variant “, but this classification complicates the reasoning of the relationship model between the processes themselves. So let’s stick with the limitation we can only classify the relationship as “connected to”.

Processes can belong to groups. You can designate groups of processes that you want and you can create groups of groups of processes.
Today business processes are grouped together according to many models:

  • According to a standard business framework: SCOR, eTOM, UBPF,ITIL;
  • According to a industry framework like airport management divided into non-aviation and aviation;
  • According to a framework defined by the company;
  • Without any specific rule.

The processes are composed by activities and tasks without it they did not exist. Processes can be structured (those who have to be repeatedly executed in the same way, often represented by flowcharts if necessary) and unstructured have a particular sequence of activities that were not previously predefined. Process groups can be grouped together. It’s possible to have groups of processes and groups of groups.

Tasks have participants (those who perform the work) and the processes have managers (someone responsible, not the classical definition of process manager) the same is true for process groups, although this condition does not always apply. For those who argue that in AD-HOC processes such definitions does not exist, there may not be formally an owner, as in traditional approach to process management, but someone oversees the progress and the outcome achievement.

In a matrix form (represented in a binary adjacency matrix), using the principles of mathematics it possible to represent algebraic relationships in this way:

The binary value 1 between two processes means that there is a relationship between them, the binary value 0 means no relationship exists between the processes.

What matters here is not discussing an abstract representation of the matrix because it is governed in accordance with the principles of algebra, and secondly the representation can be stored in a database. Therefore we will not make analysis about foundations of the theorems of mathematical representation. The important part is that you can do with it.

The creation of a true process social network, allows to build an ontology, must being able to answer such questions automatically:

  • What we do inside our organization?
  • How people collaborate?
  • What is the meaning of each logical domain in the organization?
  • What is the relationship between intra-existing processes?
  • What is the nature of the relationship between the processes?

These questions are also valid for AD-HOC processes that are created automatically the first time and relates with any object.

As we will see later this model allows to understand the relationship between the informational entities starting from the relationship between enterprise interactions.

The creation of process ontology is valueless regarding such representation. More importantly the possibility of building a process ontology, is to build a domain ontology, things, business entities, enterprise concepts.

All processes have informational entities: Customers, Complaint, Order. These entities are managed and over a certain life cycle as the processes are being executed: they are created, read, updated, and deleted. The concept and interpretation of each of these informational entities are the biggest challenge facing organizations to succeed in building a domain ontology.

On the other hand with the emergence of AD-HOC processes in which people get the informational entities that need to work and execute process instances, if there is not a clear idea about the meaning of each thing, creating barriers to “knowledge management” and “learning organizations”.

This example adapted from [14] is symptomatic of what I mean. Although the authors focused much on the issue of process modeling, representation conflicts, and explanation of the graphical representation, what matters is what is your interpretation when dealing with information, the meaning of each thing, each entity.

Semantic conflicts in the real world

To create a domain ontology is necessary to apply a set of techniques used in social network analysis. These techniques are based on the same mathematical principles of algebra as described before. I will explore it on part 3. But if you can you can move to the epilogue.


[2] Process Mashups: Helping Project Teams Put the Pieces Together.

[11] SAP Modeling Handbook –

[12] Process Redesign: The Implementation Guide for Managers – Arthur R. Tenner e Irving J. DeToro,  Prentice Hal – ISBN – 0-201633-91-4

[13] Social networks in enterprise world – POSI XII – Alberto Manuel, André Brandão, João Graça.

[14] Semantic Business Process Management – Jorg Becker, Daniel Pfeiffer, Thorsten Falk, and Michael Rackers do livro Handbook of Business Process Management, Springer, ISBN – 3-642004-15-6

[37] Tasks, Responsibilities, Practices – Peter Druker – Harper Paperbacks – ISBN 088-730-615-2

[39] The anachronism of acronyms – Gary Comerford –


6 thoughts on “Semantic BPM part two

  1. Pingback: Semantic BPM – Part One « End to End BPM

  2. Alberto, this is a good representation of current process theory. We agree on the need for an ontology to avoid ambiguity. I do however not see that Ad-hoc tasks in BPM-speak are the better description for case management or unstructured work. Ad-hoc tasks are typically not logically connected to the normal process flow. The same is true for adding a comment stream as ‘social lipstick’ to BPM. It is not an ‘activity stream’ if it does not logically interact with the process functionality, data and content. The same is true for the dependencies of process entities and they are not just tasks or activities. There can be complex any-to-any relationships that cannot be practically described in a matrix and they are certainly not helpful to a business person in any way.

    Case management is a different paradigm to BPM as it focuses already on the informational entities and their STATE. It is not the position in the process that defines next action, but the summary state of all information entities in the case context. Case management is also GOAL-oriented in difference to BPM that is output oriented. Therefore the ability for a user to add an arbitrary ad-hoc step is not enough functionality at all.

    When I talk about a Business Architecture to create an ontology I refer to not only describing the language of process but also the domain ontology of information entities. The BA is even more important to allow learning, meaning the creation and reuse of new knowledge. Without that ontology knowledge cannot be recorded. Ad-hoc processes do not in any way add new relevant knowledge unless they are linked logically as rules or patterns with their relationsships to a business goal.

    • Max, one other important thing:

      The matrix representation is a method than is easy to understand and relate objects, for this purpose only.

      The UML model included on part II is capable of dynamic, complex, recursive relationships at is was used and proved in the real world and It has no lock in. Applying the model on object social interactions enables to understand the social network of any business ecosystem, necessary to semantic discovery, evolution and management.

  3. Pingback: Semantic BPM part three « End to End BPM

  4. Pingback: Semantic BPM – epilogue « End to End BPM

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s