FINDING REUSABLE BUSINESS PROCESS MODELS

BASED ON STRUCTURAL MATCHING

Han G. Woo

Le Moyjne College, 1419 Salt Springs Road, Syracuse NY 13214, U.S.A.

Keywords: Business process models, Reuse, BPMN, Structural data mining.

Abstract: Successfully integrating business processes with information systems has been a critical issue in many

organizations. Such integrations should take place throughout the various stages of systems development to

manage correct, traceable business process requirements. To support business process management (BPM)

activities, many modeling formalisms and tools were proposed. Yet reuse of business process knowledge

has been understudied although reuse practice is common, often relying on human recollection and

reference models. This research proposes a tool support that assists reuse of business process models such as

BPMN, EPC, and UML Activity Diagrams. In the suggested approach, the semantics of these formalisms

are preserved in the conceptual graph format along with their instantiations and interrelationships. A

structural data mining tool is then used to find reusable process models based on similarities in sequences of

events and processes. This study can be applied to many reuse-related situations, namely retrieval of

reusable process models given a problem, discovery of sequence patterns among process models, and

suggesting the instances of (anti-) patterns for learning purpose.

1 INTRODUCTION

Since the management trend of business process

reengineering in the 1990s, business process

management (BPM) has been a critical issue in

many organizations (Smith and Fingar 2003). A

business process is loosely defined as a “set of

partially ordered activities intended to reach a goal

(Hammer and Champy 1993),” even though there is

a great deal of variation among its definitions

depending on which aspect of business processes is

of interest: social construct, dynamic system, or

machine metaphor (Lindsay et al. 2003). BPM

becomes more important in the fast-changing digital

economy as organizations’ business processes

themselves evolve over time to keep up with

competitive market pressure. The key aspects of

BPM are to fully comprehend an organization’s

business processes and to manage necessary changes

in these processes to meet the organization’s

strategies. For successful integration of the BPM

practices with enterprise information systems,

various business process (or workflow) modeling

formalisms and tools have been suggested (Stohr

and Zhao 2001; Weske et al. 2004).

As business process models have been analyzed

and accumulated in various projects, reuse of

business process models becomes common in

practice although it often relies on human

recollection and reference models (Thomas et al.

2006). System engineers or subject matter experts

recall business process models that were previously

constructed in a domain similar to their current

project. They search relevant business process

models from their memories or a repository of

archived documents, and then apply retrieved

models to the current context. Reusing business

process models, akin to other analysis and design

artifacts in software development, can bring various

benefits to organizations. Reusable business process

models can facilitate communications between

system engineers and clients instead of starting from

scratch, thereby help organizations capture correct

business process requirements and identify possible

improvements. In this way, they can expedite the

process of business process management to quickly

respond to changing business environments. In

addition, organizations may obtain best-practice

business processes because reusable process models

have been already validated and successfully

integrated in a similar domain.

249

G. Woo H. (2009).

FINDING REUSABLE BUSINESS PROCESS MODELS BASED ON STRUCTURAL MATCHING.

In Proceedings of the 11th International Conference on Enterprise Information Systems - Information Systems Analysis and Speciﬁcation, pages

249-255

DOI: 10.5220/0002012002490255

 SciTePress

Exploiting potential benefits of BPM appears to

have redrawn attention from researchers in recent

years. It is mainly because of (1) standardization

efforts for business process modeling languages

such as BPMN, BPEL, and XPDL (Dreiling et al.

2008) and (2) shifted interests toward Web-based,

SOA(Service-oriented architecture) business

applications (van der Aalst et al. 2007).

Interoperable business process model specifications

and packaging them as Web services render reuse of

business process not only much easier but more

attractive in that there are more reusable assets

available to organizations and ready for system

integration with slight adaptation.

Yet reuse of business process knowledge has

been understudied compared to all the progress

made in BPM research in the last two decades

(Hidders et al. 2005 ). Most studies related to

process reuse place their focus on utilizing “bigger

chunks” of business processes like process

templates, domain reference models (Thomas et al.

2006), or interoperable services (Brambilla et al.

2006; Distante et al. 2007; O'Brien et al. 2008;

Tarantilis et al. 2008) coupled with business

processes, with little attention to how to find

reusable business processes based on similarities

among actual activity and control sequences.

This research proposes a tool support that assists

reuse of business process models such as BPMN,

EPC, and UML Activity Diagrams. In the suggested

approach, the semantics of these specifications (e.g.,

event, task, sub-process, gateway, sequence,

message, and data object) are preserved in a

conceptual graph format, along with their

instantiations and interrelationships. A structural

data mining tool is then used to find reusable process

models based on similarities in sequences of events,

processes, and control structures. The structural

matching approach can complement other reuse

methods like classification based on descriptors or

attributes of business processes, and domain models.

2 RELATED WORK

Many commercial tools and academic work

implicitly or explicitly provide some reuse support

for business processes based on reusable asset

management or knowledge management perspective.

OMG’s RAS (Reusable Asset Specification)

standard (Object Management Group 2005) provides

guidelines for profiling reusable software asset.

Business process models may be managed in many

forms: requirements, artifacts, diagram, or services

(Park et al. 2007). Another approach to business

process reuse can be found in application of

reference models. MIT process handbook is an

example of such reference models that contains a

comprehensive online library of business process

knowledge (Malone et al. 2003). Thomas et al.

(2006) proposes a reference model management

system (RMMS) that facilitates development and

management of business process reference models.

Some of main ERP vendors also offer BPM tools −

for example, SAP’s NetWeaver and Oracle’s Oracle

Workflow − that provide customers with workflow

or business scenario templates and process patterns.

These templates are normally used for typical

business processes such as order processing, those

which draw on a reference model of industrial

business practice.

It appears that the increasing popularity of SOA

(Service-oriented architecture) Web applications has

also affected BPM research streams. In SOA,

business processes are bundled with service

architectures and reused as a form of context-aware

services independent of development technologies

and platforms (Brambilla et al. 2006; Distante et al.

2007; O'Brien et al. 2008; van der Aalst et al. 2007).

The commonality among these examples of

process reuse research is that the guidelines for reuse

derive from similar business context, attributes, or

descriptors, not detailed process sequences.

Therefore, little assistance is available to business

process analysts when they seek instance-level

exemplars that can be applied to generate alternative

business processes or to manage dynamic changes in

existing processes.

A few studies in workflow management system

(WfMS) research community attempt to assist in

workflow reuse based on workflow sequences and

control structures. van der Aalst et al. (2003)

suggests generic workflow patterns. These domain-

independent patterns were initially defined using

control flows; the patterns have evolved over time

including the observations of various perspectives in

workflows: data, resource, and exception handling

(N. Russell et al. 2006). Some of the basic control-

flow patterns are used in this study to illustrate

business process queries.

The process mining tool such as ProM (van der

Aalst 2007) aims to discover reusable business

processes from a-posteriori analysis of event logs

that record activities, timestamps, roles, and related

data object. It claims that the discovered patterns

from process mining are more practical and realistic

because it looks at “inside the process” at a very

refined level.

ICEIS 2009 - International Conference on Enterprise Information Systems

250

Madhusudan et al.’s work (2004) employs case-

based reasoning to support workflow modeling and

design. Their framework deals with business process

model management issues, from storage, retrieval, to

reuse and adaptation. To find reusable workflow

cases, a similarity-based case retrieval method called

Similarity Flooding algorithm is used. This

technique shares some characteristics with the tool

support presented in this paper in that it finds

matching two (query and source) directed graphs

based on semantic similarity in node and edge labels

and similarity in topology of the graphs. However

the representation of business process models and

similarity metric in Similarity Flooding are rather

simpler; it does not consider detailed process

elements expressed in standard modeling

specifications.

3 STRUCTURAL MATCHING

This section explains an approach to reuse of

business process models based on structural

similarity. First, a system engineer or reuse

administrator defines or collects business process

models in BPMN, EPC, or UML Activity diagram

notations. These models are transformed into

directed conceptual graphs that consist of vertices

and edges. The conceptual graphs are added into a

business process library along with the initial

models. As an analyst begins defining a business

process model or tries to find applications of a

certain pattern, the analyst can call on the tool

implemented with the data mining algorithm. The

algorithm searches for similar structures in the

library and returns best matches. The analyst then

selects the most relevant business process model and

adapts it to the current analysis problem.

3.1 Structural Data Mining

The tool support suggested in this paper employs an

automated relational learner called Subdue

(Gonzalez et al. 2000; Joyner et al. 2001) that

discovers patterns in structured data sets. In Subdue,

information is stored as a graph of vertices and

edges. Vertices usually refer to objects, attributes,

and their values while edges represent relationships

between the objects. The syntax for vertex

description is <v id label> where id is a vertex

number and label is the name of that vertex. Edges

are coded with <u id1 id2 label> or <d id1 id2

label>. The former represents an undirected edge (u)

between vertex id1 and id2; the latter means a

directed edge (d) from vertex id1 to id2. Examples

of the Subdue graph are shown in Figure 1.c and

Table 1.

Subdue’s search algorithm finds repetitive

substructures called concepts in graphs. The search

starts with a uniquely labeled vertex of a graph

initializing the search queue. Following the beam

search strategy, Subdue expands its search by

including adjacent edges and associated vertex in all

possible ways, yielding potential substructures.

When a repeating substructure is found, it is

replaced with a placeholder vertex pointer to its

substructure, thereby compressing the whole graph.

Each candidate substructure is evaluated by a

compression score. The compression score is

calculated by (DL(S) + DL(G|S)) / DL(G) where

DL(G) stands for the description length of the input

graph, DL(S) the description length of the

substructure, and DL(G|S) the description length of

the input graph when compressed by the

substructure. This evaluation metric bases its

assumption on the Minimum Description Length

(MDL) principle. It states that the best concept

(substructure) describes the whole data set with a

minimal description length, i.e. the length in number

of bits of the graph representation when compressed

by the substructure (Cook and Holder 2000). When a

candidate substructure is found better than others in

terms of the compression ability, it is stored in the

best substructure queue. Iterating this process results

in a hierarchical classification lattice whose lower-

level concepts are included in the higher-level

concepts. The iteration can be limited by two

parameters: breadth of search (beam) and number of

expansions (limit.) The search terminates when it

reaches a user specified limit on the number of

substructures extended or when the search space is

exhausted.

In Subdue, there are two important features

relevant to our purpose. One is its ability to find

inexact match using threshold value. It is especially

useful because finding reusable business processes

normally requires a certain degree of tolerance in

their variations caused by different styles in

authoring models and by differences in naming the

same concept. In Subdue, a threshold value

determines when two structures are similar enough

to match. The analyst can set this threshold

parameter from 0.0 to 1.0. The value 0.0 means a

complete match and 1.0 the maximum tolerance

level. The similarity metric of two structures is

computed as transformation cost / structure size,

where transformation cost is the number of graph

transformations required to make the structures

FINDING REUSABLE BUSINESS PROCESS MODELS BASED ON STRUCTURAL MATCHING

251

isomorphic. Two structures match when the

similarity metric is less than the threshold (Cook and

Holder 2000). The other feature is the capability that

deals with synonyms. A list of predefined synonyms

can substitute different vertex or edge labels that

carry the same meaning. This functionality enables

Subdue’s potential to utilize the benefits of a domain

ontology or lexicon.

3.2 Representation of Business Process

Models

As described in the previous section, business

process models need to be represented as directed

conceptual graphs in order for the structural data

mining algorithm to find a similar match. The

transformation of business process models into

conceptual graphs takes place at two different

abstraction levels: metamodel and instance level.

Since key elements in metamodel types in BPMN,

EPC models, and UML activity diagrams share

similar semantics, converted process graphs can be

used together regardless of differences in the

modeling notations.

Let us consider an example of BPMN models.

BPMN core semantics include swimlanes, events,

activities, gateways, and other artifacts (Object

Management Group 2008). Table 1 summarizes the

representation scheme for BPMN notations. These

metamodel elements are coded with vertices in a

conceptual graph. Each vertex has a label like Pool,

Lane, StartEvent, IntermediateEvent, EndEvent,

Task, SubProcess, Gateway, DataObject, etc. The

connecting objects such as sequence flows and

message connectors are coded as edges between two

vertices that represent these metamodel elements. A

pool or data object instance also becomes a vertex

with its name as a vertex label. Then the instance

and its metamodel vertex are connected with a

directed edge labeled InstanceOf. A task instance

has two vertices describing the nature of the task

with a verb and an object. They are linked with

ActionOf and ActionObjectFor edges respectively.

Gateway instances are coded in a similar fashion.

For example, if there is an exclusive (XOR) gateway

with two branches, a gateway name becomes a

vertex pointing to the metamodel vertex Gateway

with an edge ConditionOf. The two branches are

represented as vertices and connected with edges,

DefaultBranchOf or BranchOf. Figure 1 illustrates a

detailed example of the transformation. Figure 1.a

shows a fragment of the BPMN model, Process

Order while Figure 1.b illustrates the translated

vertices and edges in the conceptual graph format. In

Figure 1.b, the gray-highlighted part represents the

structural information of the process, and the white-

colored elements are the instance-level information.

Figure 1.c is the actual Subdue text graph used in the

algorithm.

The transformation coding scheme is designed

with two guidelines: (1) separation between

metamodel elements and instances and (2)

maintaining an atomic value for each vertex and

edge. These guidelines allow the search algorithm to

focus more on structural aspects of business process

models and to handle naming differences by

populating synonyms. This way, an intermediate

language between natural language and the formality

of first-order logic makes it possible to perform

classification, aggregation, and generalization

(Greenspan and Mylopoulos 1982).

Table 1: Transformation of BPMN into Subdue Graph.

BPMN Semantics

BPMN Example Graph Example

Pool

V 1 Pool

V 2 Sales

D 2 1 InstanceOf

Lane

V 1 Pool

…

V 3 Lane

V 4 Sales Rep

D 4 3 InstanceOf

D 3 1 Within

MessageStartEven

V 1 StartEvent

V 2 Message

V 3 Credit Request

D 2 1 TypeOf

D 3 2 newStateOf

Task

V 1 Task

V 2 Evaluate

V 3 Credit

D 2 1 ActionOf

D 3 2 ActionObjectFor

Gatewa

V 1 Gatewa

V 2 Approved?

V 3 Exclusive

V 4 Yes

V 5 No

D 2 1 ConditionOf

D 3 1 TypeOf

D 4 1

DefaultBranchOf

D 5 1 BranchOf

Sequence Flow

V 1 Task

V 2 Send

V 3 RFQ

D 2 1 ActionOf

D 3 2 ActionObjectFor

V 4 Task

V 5 Receive

V 6 Quote

D 2 1 ActionOf

D 3 2 ActionObjectFor

D 1 4 Sequence

ICEIS 2009 - International Conference on Enterprise Information Systems

252

a. BPMN model

% Pool

V 1 Pool

V 2 Sales

D 2 1 InstanceOf

% Event

V 3 StartEvent

V 4 Empty

D 4 3 TypeOf

% Task (Receive Order)

V 5 Task

V 6 Receive

V 7 Order

D 6 5 ActionOf

D 7 6 ActionObjectFor

% Gateway

V 8 Gateway

V 9 Exclusive

V 10 Accepted?

V 11 Yes

V 12 No

D 9 8 TypeOf

D 10 8 ConditionOf

D 11 8 DefaultBranchOf

D 12 8 BranchOf

% Task (Close Order)

V 13 Task

V 14 Close

V 15 Order

D 14 13 ActionOf

D 15 14 ActionObjectFo

…

% Sequence Flow

D 1 3 Initiate

D 3 5 Sequence

D 5 8 Sequence

D 11 13 Sequence

…

c. Graph Text

Pool StartEvent

Sales

Gateway

InstanceOf

Empty

TypeOf

Sequence

Exclusive

TypeOf

Accepted?

Default

BranchOf

Task

Receive Order

ActionOf

Action

ObjectFor

Sequence

Yes

ConditionOf

BranchOf

Task

Close Order

ActionOf

Fill Order

Task

ActionOf

Sequence

Action

ObjectFor

Action

ObjectFor

b. Subdue Graph

Figure 1: Example of Transformation from BPMN to Subdue Graph.

4 PRELIMINARY EVALUATION

To demonstrate the feasibility of the tool support

suggested in the paper, a simple evaluation was

taken with a relatively small process library. It

consists of 37 partially or fully completed business

process models in BPMN notations, yielding 1089

vertices and 1151 edges. The business process

models were collected from the specification

documents, examples available on related Web sites,

and textbooks.

Since the main purpose of this case study is to

see whether the tool can assist in finding similar

business process structures, seven queries were

presented to the library. The queries are borrowed

from the workflow patterns in (van der Aalst et al.

2003), specifically Sequence, Parallel Split,

Synchronization, Exclusive Choice, Simple Merge,

and N out of M Join. These query models are shown

in Figure 3. In addition, the model in Figure 2.a (part

of order processing) is included for a more

complicated query.

For each query, the threshold value is initially set

as 0.0 and then incremented by 0.1 until 0.9. Table 2

summarizes the retrieval results for the queries. In

each trial, relevant, best matched graphs were

retrieved at the threshold between 0.3 and 0.9. For

simple queries like Sequence, Parallel Split, and

Synchronization, the tool was able to find similar

business process models at the relatively low

threshold values. Since the query graphs used simple

labels such as A, B, B1, etc., there was no exact

match found. In the first query Sequence, there are

too many instances found at the threshold > 0.7,

suggesting almost all sequential chains in the library.

For the complicated queries, N out of M Join query

fins only one match at 0.8. The retrieved process

fragment contains an exclusive gateway instead of a

complex gateway. For Process Order query, the

completed business process model was intentionally

prepopulated with a few modifications on vertex and

edge labels, and it was found at 0.8.

The results of the preliminary evaluation suggest

that structural matching technique can be applied to

find relevant, reusable business process models. Yet,

in order for the tool to be practical, each search must

be tuned with a proper threshold. It should be also

noted that for complicated queries with a high

threshold, the computation time may exceed more

than a minute in a personal computer CPU

environment because the algorithm itself is

polynomial. This concern can be resolved by

adjusting other search options such as limit, beam,

number of vertices in a structure, etc.

FINDING REUSABLE BUSINESS PROCESS MODELS BASED ON STRUCTURAL MATCHING

253

Sequence

Synchronization

Simple Merge

Parallel Split

Exclusive Choice

N out of M Join

Figure 2: Workflow Patterns used as Queries.

Table 2: Query Results.

Query Best match found

at threshold

Number of

Instances

Sequence 0.3 – 0.6 > 45

Parallel Split 0.3 – 0.7 > 23

Synchronization 0.3 – 0.7 > 7

Exclusive Choice 0.4 – 0.7 > 11

Simple Merge 0.6 – 0.8 > 9

N out of M Join 0.8 – 0.9 1

Process Order 0.7 – 0.9 1

5 DISCUSSION

This research proposes automated tool support for

business process reuse that exploits rich semantics of

business process modeling formalisms. Business

process models are translated as conceptual graphs

that comprise vertices and edges. The coding

scheme is quite flexible and extensible; it can

express core semantics of existing business process

specifications. By applying the structural matching

technique, the tool support can deal with a certain

degree of informality inherent in business process

models while looking at similar sequence patterns.

This study can be applied to many reuse-related

situations, namely retrieval of reusable process

models given a problem, uncovering sequence

patterns among process models, and suggesting the

instances of (anti-) patterns for learning purpose.

Future work includes developing the prototype of

the tool support, validating its effectiveness in a field

or lab experiment setting. The prototype of the tool

support is under development as a plug-in to SOA

Tools Platform on Eclipse (http://www.eclipse.org/

stp/bpmn.) This approach is being tested with a

bigger collection of business process models for

ERP systems in order to support ERP configuration

with business process modeling. Additional methods

for search tuning also need to be explored to

increase search performance, including ontological

support of important concept matching and Subdue’s

supervised learning feature.

REFERENCES

Moore, R., Lopes, J., 1999. Paper templates. In

TEMPLATE’06, 1st International Conference on

Template Production. INSTICC Press.

Smith, J., 1998. The book, The publishing company.

London, 2

edition.

Brambilla, M., Ceri, S., Fraternali, P., and Manolescu, I.

"Process modeling in Web applications," ACM

Transactions on Software Engineering Methodologies.

(15:4) 2006, pp.360-409.

Cook, D.J., and Holder, L.B. "Graph-Based Data Mining,"

IEEE Intelligent Systems (15:2) 2000, pp.32-41.

Distante, D., Rossi, G., and Canfora, G. "Modeling

business processes in web applications: an analysis

framework," in Proceedings of the 2007 ACM

symposium on Applied computing, ACM, Seoul,

Korea, 2007, pp.1677-1682.

Dreiling, A., Rosemann, M., van der Aalst, W.M.P., and

Sadiq, W. "From conceptual process models to

running systems: A holistic approach for the

configuration of enterprise system processes,"

ICEIS 2009 - International Conference on Enterprise Information Systems

254

Decision Support Systems (45:2), 2008/5 2008,

pp.189-207.

Gonzalez, J.A., Jonyer, I., Holder, L.B., and Cook, D.J.

"Efficient Mining of Graph-Based Data," Proceedings

of the AAAI Workshop on Learning Statistical Models

from Relational Data, 2000, pp.21-28.

Greenspan, S.J., and Mylopoulos, J. "Capturing More

World Knowledge in the Requirements Specification,"

Proceedings of the Proceedings of the 6th

International Conference on Software Engineering,,

IEEE CS Press, Los Alamitos, CA, 1982, pp.225-234.

Hammer, M., and Champy, J. Re-engineering the

Corporation; A Manifesto for Business Revolution,

Harper Business, New York, 1993.

Hidders, J., Dumas, M., Aalst, W.M.P.v.d., Hofstede,

A.H.M.t., and Verelst, J. "When are two workflows

the same?" in Proceedings of the 2005 Australasian

symposium on Theory of computing, Australian

Computer Society, Inc., Newcastle, Australia, 2005

pp.3-11

Joyner, I., Cook, D.J., and Holder, L.B. "Discovery and

Evaluation of Graph-Base Hierarchical Conceptual

Clusters," Journal of Machine Learning Research (2)

2001, pp.19-43.

Lindsay, A., Downs, D., and Lunn, K. "Business

processes--attempts to find a definition," Information

and Software Technology (45:15), 2003, pp.1015-

1019.

Malone, T.W., Crowston, K., and Herman, G.A. (eds.)

Organizing Business Knowledge: The MIT Process

Handbook. MIT Press, Cambridge, MA, 2003.

N. Russell, A.H.M., ter Hofstede, W.M.P., Aalst, v.d., and

Mulyar, N. "Workflow Control-Flow Patterns: A

Revised View.," BPM Center Report BPM-06-22,

BPMcenter.org, 2006.

Object Management Group, Reusable Asset Specification,

V 2.2, 2005.

Object Management Group, Business Process Modeling

Notation, V1.1, 2008.

O'Brien, L., Brebner, P., and Gray, J. "Business

transformation to SOA: aspects of the migration and

performance and QoS issues," in Proceedings of the

2nd international workshop on Systems development

in SOA environments, ACM, Leipzig, Germany, 2008,

pp.35-40.

Park, S., Park, S., and Sugumaran, V. "Extending reusable

asset specification to improve software reuse," in

Proceedings of the 2007 ACM symposium on Applied

computing, ACM, Seoul, Korea, 2007, pp.1473-1478.

Smith, H., and Fingar, P. Business Process Management:

The Third Wave, Meghan-Kiffer Press, Tampa, FL,

2003.

Stohr, E.A., and Zhao, J.L. "Workflow Automation:

Overview and Research Issues," Information Systems

Frontiers (3:3), 2001, pp.281-296.

Tarantilis, C.D., Kiranoudis, C.T., and Theodorakopoulos,

N.D. "A Web-based ERP system for business services

and supply chain management: Application to real-

world process scheduling," European Journal of

Operational Research (187:3), 2008, pp.1310-1326.

Thomas, O., Horiuchi, M., and Tanaka, M. "Towards a

reference model management system for business

engineering," in Proceedings of the 2006 ACM

symposium on Applied computing

, ACM, Dijon,

France, 2006, pp.1524-1531.

van der Aalst, W.M.P. "Trends in Business Process

Analysis: From Verification to Process Mining,"

Proceedings of the the 9th International Conference

on Enterprise Information Systems (ICEIS 2007),

Medeira, Portugal, 2007, pp.12-22.

van der Aalst, W.M.P., Benatallah, B., Casati, F., Curbera,

F., and Verbeek, E. "Business process management:

Where business processes and web services meet,"

Data & Knowledge Engineering (61:1), 2007/4 2007,

pp.1-5.

van der Aalst, W.M.P., ter Hofstede, A.H.M.,

Kiepuszewski, B., and Barros, A.P. "Workflow

Patterns," Distributed and Parallel Databases (14:3),

2003, pp.5-51.

Weske, M., van der Aalst, W.M.P., and Verbeek, H.M.W.

"Advances in Business Process Management," Data &

Knowledge Engineering (50:1) 2004, pp.1-8.

FINDING REUSABLE BUSINESS PROCESS MODELS BASED ON STRUCTURAL MATCHING

255