Bridging the Gap between XPDL and Situation Calculus:

A Hybrid Approach for Business Process Verification

Bing Li and Junichi Iijima

Graduate School of Decision Science and Technology

Tokyo Institute of Technology, Japan

Abstract. Business Process Verification (BPV) is increasingly important in

emerging BPM Systems. Although many approaches related to BPV exist, the

gap between formal models defined in previous research and informal models

used in the industry prevents these approaches from large-scale industrial ap-

plications. This paper proposes a hybrid approach to bridge the gap between

XPDL and Situation Calculus. XPDL defines a business process informally

while Situation Calculus provides a formal specification to enable the function

of BPV. The gap between them will be bridged by a devised language – XSSL,

based on which the process model will be logically verified.

1 Introduction

Evolving from traditional workflow management and embracing some new comput-

ing concepts, emerging business process management systems (BPMS) are trying to

make strength on system integration [1, 2]. As part of the BPMS, the business process

verification (BPV) emphasizes the correctness properties of business processes [3].

Despite the increasing importance of BPV in the BPMS, the verification function is

still not sufficient [4]. Presently many BPM tools have no explicit or strong verifica-

tion about process definitions.

Research related to BPV exists, in some of which formal methods are applied to

verify business processes and thus a process model can be ensured to some degree of

robustness. Due to an emphasis on the underlying mathematical foundation, these

approaches are not yet well applied to the BPM industry. The reason resides in the

existing gap between the formal models defined in these researches and the informal

models usually used in the industry. Explicit connections with industrial standards or

transformation rules from them to formal models are necessary in order to bridge or

decrease the gap between them.

Flexibility is also an important concern in BPMS. In order to improve the integrat-

ability of the BPMS, the industry is making great efforts in standardization. The BPM

related research would also benefit from sharing some industrial standards such as

business process specification languages. This will enable collaboration among dif-

ferent verification methods and thus reach a higher degree of robustness.

This paper tries to solve the problem in the face of BPMS by concentrating on

BPV. A hybrid approach is proposed to verify the XPDL-based process definition.

Li B. and Iijima J. (2007).

Bridging the Gap between XPDL and Situation Calculus: A Hybrid Approach for Business Process Veriﬁcation.

In Proceedings of the 5th International Workshop on Modelling, Simulation, Veriﬁcation and Validation of Enterprise Information Systems, pages

151-156

DOI: 10.5220/0002416501510156

 SciTePress

The industrial standard – XML Process Definition Language (XPDL) – will be used

to describe the source process model while Situation Calculus works as the underly-

ing logical language to make verification. For convenient transformation from XPDL

to Situation Calculus and improvement in reusability of process or activity specifica-

tion, XML Situation-calculus Specification Language (XSSL) is originally proposed.

The XSSL tries to bridge the gap between the informal and formal process models.

This paper is organized into four sections. Section 2 briefly reviews the related re-

search. Then the BPV approach is explained in detail in Section 3. Finally the conclu-

sion and future work are given in Section 4.

2 Related Work

Business Process Verification (BPV) is to verify the correctness of the business proc-

ess definition. It usually applies a formal method and specific correctness properties

are defined and verified according to the employed formalism. A formal process

model does not only make mathematical verification possible, but also enables poten-

tial extension for intelligent functions [4, 5]. There exist many formal approaches

related to BPV, in which Petri-Net based and logic based approaches are two popular

categories.

Petri-Net has been widely used and extended ever since it was invented. The Petri-

Net based approaches for business process modelling have some obvious advantages:

one is the consistence with the task-oriented view of business processes; a second

merit is the graphical notation. These make it popular in the field of workflow model-

ling. The critical problems in these approaches include the expressive power such as

capability of data modelling and in some degree they have been solved by some ex-

tension of the classic Petri-Net, such as including hierarchical or object-oriented con-

cepts [6, 7].

Logic-based approaches form another category. Usually mathematical logics are

the theoretical foundation of these approaches. By using a specific logic language,

business processes are precisely specified. Based on the formal specification, some

proof tasks including verification can be performed. The reasoning function distin-

guishes these logic-based approaches from other ones, which can make the business

system more intelligent and dynamic. But since these approaches emphasize the logic

notation and calculi, they are not easily understood by the business process analysts.

Moreover it takes efforts to formally define business processes [8, 9, 10].

Generally speaking, the present BPV approaches are in need of improvement in

order to be applied to the BPM industry. Connections with the emerging industrial

standards are necessary and the gap between the industrial and academic process

models should be bridged. This paper tries to provide a hybrid approach to integrate

such informal and formal process models. This way the separate strengths such as

practicability and robustness can both be obtained.

152

3 BPV Approach

In order to bridge the gap between the informal and formal process models, a hybrid

approach for BPV is proposed in this paper. Firstly we will explain this approach

from the perspective of model transformation in the desired system.

3.1 Model Transformation

In this approach there are three forms of process models: XPDL, XSSL and Prolog.

XPDL is an industrial standard and supported by many BPMS vendors. In this ap-

proach XPDL is selected as the source process model for its analyzability in XML

syntax. XPDL focuses on the business logics and can specify the transition relations

in business processes. Constituents in a process are separately represented as Work-

flowProcess, Activity, Transition and so on [11]. This paper concentrates on the con-

trol flow perspective, that is, transitional relations in business processes. But this

approach is possibly applied to verify other aspects of business processes such as

global constraints [8].

XSSL is designed based on some concepts in Situation Calculus. The direct pur-

pose to design this XML-based language is to bridge the gap between the XPDL

defined process model and the logic-based process model. Thus transformation be-

tween them is more convenient. Moreover, XSSL will enable separation of activity or

primitive process specification from complex processes and improve their reusability.

Furthermore, the XML-based syntax of XSSL will enable potential application of

Situation Calculus in web service computing such as service choreography.

Prolog works as the implementation language for logical reasoning. It will repre-

sent a process model logically by using Situation Calculus. Based on this inferable

model, the business process will be formally verified. Benefiting from this formal

model, other intelligent analysis will be possibly implemented such as planning or

monitoring in BPMS.

Between these process models there are two types of transformation. The first

transformation is from XPDL to XSSL. Information about processes, activities, tran-

sitions and conditions will be extracted from an XPDL file, and automatically trans-

formed to the compact format in XSSL. The second transformation is from XSSL to

Prolog. This will automatically transform an XSSL specified process model to a

Prolog program.

3.2 Theories about Transformation

Formal definition about the approach is being written in [13] which concentrates on

the underlying formalization of XPDL, XSSL and the mapping between them by

employing mathematical structures. In this paper we focus on the overall introduction

about the approach.

153

3.2.1 Situation Calculus

Situation Calculus has strength of reasoning about actions. This formalism can be

applied to BPV [9]. The semantic transformation from XPDL to Situation Calculus

seems intuitive and uncomplicated. Furthermore Situation Calculus is extensible to

include some dynamic features such as concurrency and reactiveness [9, 12].

3.2.2 XSSL

XSSL is initially designed to contain the compact format of action specification in

XML syntax. It includes some basic elements such as actions, arguments, precondi-

tions and postconditions. The following is part of an XSSL file that is generated

automatically by parsing the XPDL file.

<args>

<arg>CardNo</arg>

</args>

OrderStatus=="received"

</precondition>

PayWay=="credit"

</precondition>

</preconditions>

CreditStatus=="valid"

</postcondition>

OrderStatus=="checked"

</postcondition>

</postconditions>

</Action>

The defined elements in XSSL correspond intuitively to the concepts in Situation

Calculus except for that postconditions refer to successor state axioms. Compared

with XPDL, the Action element corresponds to an activity in XPDL. The args ele-

ment represents the parameters of an activity. The preconditions and postconditions

are transformed from the conditions in XPDL by using some defined rules for differ-

ent transition types [13].

From XSSL, the Prolog formatted specification about actions involved in a process

can be automatically generated. Some additional conditions will be added by restrict-

ing the relation between arguments.

3.2.3 Business Process Verification

Based on the formalism of Situation Calculus, the BPV is defined by the following

three properties for actions:

• Executability. For an action, it is possible to be executed.

154

• Satisfiability. For an action, its postcondition can be satisfied.

• Effectiveness. For an action, a satisfiable situation is effective.

Executability ensures that the precondition of an action is satisfied, i.e., the action

can be executed; Satisfiability states that the postcondition of an action can be satis-

fied; Effectiveness ensures that the satisfiable situation is effective, that is, this situa-

tion is the successor state resulting from executing the action at the executable situa-

tion. Stronger effectiveness is interpreted as that the satisfiable situation is also in the

set of possible action sequences that can be extracted from the XPDL routing infor-

mation.

From the above properties, it can be seen that a business process is obviously veri-

fied by its actions (activities). A process consists of a collection of coordinated activi-

ties. The pre-and postcondition of each activity in the context of a process also in-

clude some constraints from the process structure. After the three properties are veri-

fied for each action, the whole process is accordingly ensured to be executed and

reach the goal situation.

3.2.4 The Prototype System

A prototype system is designed and developed to demonstrate the feasibility of this

BPV approach. The XPDL defined process definition is the input of this BPV system.

The XPDL file is automatically transformed to XSSL; and XSSL can be automati-

cally transformed into a Prolog file. Then the user can verify each activity in a proc-

ess by interacting with the system such as setting initial situation or modifying transi-

tion conditions.

3.2.5 Related Issues

Some related issues are necessary to explain and clarify the approach. This approach

is concentrating on the control flow perspective of a business process. Transition

relations in XPDL are verified with Situation Calculus. But this approach can be

extended to other aspects of a business process.

The current verification focuses on the level of a workflow process. But when a

complex business process consisting of several workflow processes is considered as a

procedure or complex action, the verification can be extended to check complex busi-

ness processes. Golog is just such a programming language based on Situation Calcu-

lus. During the verification, we also use a Golog interpreter [12] and thus this work

can be directly extended to complex processes.

Furthermore, the possible transition routes can be extracted from XPDL. By pars-

ing such information, the possible action sequences or situations in Situation Calculus

can be obtained. This can be used to aid verification and improve Prolog performance

by avoiding some backtracking on situations. This aspect will be explicitly defined in

our next work.

155

4 Conclusions

This paper proposes a hybrid approach for business process verification. The indus-

trial standard XPDL and the formalism of Situation Calculus are selected as the lan-

guages for process specification. Although they are usually used in different fields,

they share some functions in specifying processes or activities. By linking them to

cooperate in business process verification, some meaningful results will be obtained.

The practicability and robustness are direct merits. Besides, the target model is infer-

able and can enable more intelligent functions.

XSSL is devised to bridge the gap between XPDL and Situation Calculus. Some

concepts of Situation Calculus are represented in XSSL with the syntax of XML

while encapsulating the underlying logic calculus. This attempts to make part of this

formalism accessible to general business analysts and furthermore enable potential

application in service computing.

There is still much work to be done in our future research and some extension work

was explained in Section 3.2.5. Other meaningful work will include the extension of

the underlying formalization of XPDL and the mapping from XPDL to XSSL. Fur-

thermore, representation of conditions in XPDL will be also necessary to improve the

degree of semantic verification.

References

1. Havey, M.: Essential Business Process Modeling. O’Reilly, 1

edition (2005)

2. Marinescu, D.C.: Internet-Based Workﬂow Management: Towards a Semantic Web.

Wiley-Interscience, New York (2002)

3. Van Der Aalst, W.M.P., Ter Hofsede, A.H.M.: Verification Of Workflow Task Structures:

A Petri-Net-Based Approach. Information System, Vol. 25, No. 1, 43–69 (2000)

4. Van Der Aalst, W.M.P., Ter Hofstede, A.H.M., Weske, M.: Business Process Management:

A Survey. BPM 2003, LNCS 2678. Springer (2003) 1–12

5. Schumann, J. M.: Automated theorem proving in software engineering. Springer (2001)

6. Reijers, H.A.: Design and Control of Workflow Processes: Business Process Management

for the Service Industry. LNCS 2617, Springer (2003)

7. Janssens, G.K., Verelst, J. and Weyn, B.: Techniques for Modelling Workflows and Their

Support of Reuse. Business Process Management, LNCS1806, Springer (2000) 1–15

8. Mukherjee, S., Davulcu, H., Kifer, M., Senkul, P. and Yang, G.: Logic Based Approaches

to Workflow Modelling and Verification. Logics for Emerging Applications of Databases,

Springer (2003)

9. Koubarakisa, M., Plexousakis, D.: A formal framework for business process modelling and

design. Information Systems 27 (2002) 299–319

10. Iijima, J.: Formal Approach for Business Process Modeling. Proceedings of National Con-

ference of Japan Society for Management Information (1999) 151–154

11. WfMC: Process Definition Interface – XML Process Definition Language. WFMC-TC-

1025, Version 2.0 (2005)

12. Reiter, R.: Knowledge in Action: Logical Foundations for Specifying and Implementing

Dynamical Systems. MIT Press (2001)

13. Li, B., Iijima, J.: Formal Verification of XPDL-based Business Process Definition. Submit-

ted to the International Journal of Business Process Integration and Management (2007)

156