An Active Rule Base Simulator
based on Petri Nets
Joselito Medina-Mar
´
ın and Xiaoou Li
1
Secci
´
on de Computaci
´
on, Departamento de Ingenier
´
ıa El
´
ectrica
2
CINVESTAV-IPN, Av. IPN No. 2508, C.P. 07360, Mexico City, M
´
exico
Abstract. Event-condition-action rules, in active database systems, should be
performed carefully, because their firings can produce inconsistent states in data-
base systems. In this paper, an ECA rule base simulator is described, named
ECAPNSim, which uses a Conditional Colored Petri Net model to depict ECA
rules. It can model ECA rules, simulate their behavior, and perform a static analy-
sis.
Keywords: ECA rule, active database system, Petri nets.
1 Introduction
Active behavior of a database system (DBS) can be defined through a base of active
rules. The model most widely used is the event-condition-action (ECA) rule model,
whose general form is:
on event e
1
if condition c
1
then action a
1
This model works in the following way: when an event e
1
occurs, if condition c
1
is
evaluated, then action a
1
is executed inside the database (DB).
Events in ADBs can be of two types: primitive or composite. A primitive event is
generated by the execution of an operation over the DB information (insert, delete, up-
date, or select), a transaction, a clock event (which can be absolute, relative, or periodic),
or the occurrence of a DB external event. On the other hand, composite events (disjunc-
tion, conjunction, sequence, closure, times, negation, last, simultaneous, and any) are
formed by the occurrence of a combination of primitive and/or composite events.
In this work, an ECA rule base simulator is presented, which is named ECAPNSim.
In order to model an ECA rule base, ECAPNSim uses a Petri Net (PN) approach, which
is based in a PN extension named Conditional Colored Petri Net (CCPN). CCPN pro-
vides properties of ECA rules, therefore, ECA rule bases can be converted into a CCPN
and then, ECA rule base simulation can be performed by using ECAPNSim.
2 Conditional Colored Petri Net
Petri Nets are a graphical and mathematical tool for modeling concurrent, asynchro-
nous, distributed, parallel, nondeterministic, and/or stochastic systems. Petri net may
be extended widely and applied in every area with logic relations. Up to now, few re-
searches have adopted Petri nets as ECA rule specification language [1], [2]. SAMOS is
Medina-Mauın J. and Li X. (2005).
An Active Rule Base Simulator based on Petri Nets.
In Proceedings of the 3rd International Workshop on Modelling, Simulation, Verification and Validation of Enterprise Information Systems, pages 96-101
DOI: 10.5220/0002566500960101
Copyright
c
 SciTePress

a successful ADB system, which partially uses Petri nets for composite event detection
and termination analysis. But, the framework is not Petri-net-based [3]. Conditional
Colored Petri Net (CCPN) is an extended PN model, which was modified in order to
support features of ECA rules [4]. In a CCPN it is very easy to detect the existence of
both relationships and dependencies between two or more rules according to its graph-
ics representation. Moreover, both primitive and composite events can be modeled.
2.1 General description
In a CCPN, ECA rule event e is stored as a place p
1
, conditional part c is stored inside
a transition t, and the action rule a, because of its similarity to an event, is stored in a
place p
2
. Therefore, if t is the transition where is stored the condition of rule r, then
•
t = {p
1
}, and t
•
= {p
2
}, where
•
t is the set of the input places of t, and t
•
is the set
of the output places of t.
During CCPN execution, events occurring inside DB are detected by the CCPN,
and if there is a CCPN place p
1
representing to detected event e then a token is gen-
erated with information about e (e.g. the record of an employee) and with a timestamp
according to the time when the event was raised. By CCPN execution, the new token is
sent to transition t
1
,
•
t
1
= {p
1
}, and the condition c stored in t
1
is evaluated against
token information. If token information is not enough to evaluate c then a query to DB
is executed to know the DB state and perform the evaluation to c. If c is evaluated to
true then one token with information about the rule action a is generated and it is sent
to place p
2
, t
•
1
= {p
2
}, which represents the ECA rule action a.
Composite events, that deal with time interval, evaluate token timestamp, and if it
belongs to composite event interval, the token is sent to its corresponding transition.
Formally, a CCPN is defined as follows [5]:
Definition 1. A conditional colored Petri net (CCPN) is a 11-tuple
CCP N = {Σ, P, T, A, N, C, Con, Action, D, τ, I}
where
(i) Σ is a finite set of non-empty types, called color sets.
(ii) P is a finite set of places. P is divided into subsets, i.e., P = P
prim
∪ P
comp
∪
P
virtual
∪ P
copy
, where P
prim
represents primitive events and it is depicted graphi-
cally as a single circle. P
comp
represents composite events negation, sequence, closure,
last, history, and simultaneous and it is depicted graphically as a double circle. P
virtual
represents composite events conjunction, disjunction, and any and it is depicted graph-
ically as a single dashed circle. And, P
copy
, is a set which is used when two or more
rules are triggered by the same event and it is depicted graphically as a double circle
where the interior circle is a dashed one.
(iii) T is a finite set of transitions. T = T
rule
∪ T
copy
∪ T
comp
., where T
rule
repre-
sents the set of rule type transitions and it is depicted graphically as a rectangle. T
copy
is the set of copy type transitions and it is depicted graphically as a single bar. T
comp
is
the set of composite type transitions and it is depicted graphically as a double bar.
(iv) A is a finite set of arcs.
(v) N is a node function. It is defined from A to P × T ∪ T × P .
97

(vi) C is a color function. It is defined from P to Σ.
(vii) Con is a condition function. It evaluates either the rule condition if t ∈ T
rule
or it evaluates the time interval when t ∈ T
comp
.
(viii) Action is an action function. It creates tokens according to action rules.
(ix) D is a time interval function.
(x) τ is a timestamp function.
(xi) I is an initialization function.
2.2 CCPN execution
CCPN execution is modified because CCPN token takes time information and CCPN
transitions are evaluated only in certain intervals, furthermore transitions t ∈ T
rule
are
evaluated with the conditional part of ECA rule. [5] In CCPN, composite transitions and
rule transitions fire conditionally. A composite (rule) transition fires once it is enabled
and the temporal (rule) condition is satisfied.
Definition 2. In CCPN, a token element is a triple (p, c, stamp) where p ∈ P , c ∈ C(p),
and stamp specifies the natural time when the token is deposited into place p. The sets
of all markings is denoted by M.
Definition 3. Transition t ∈ T is enabled at a marking M iff
1). ∀p ∈
•
t :| M (p) |= 0, type(t) = Negation
2). ∀p ∈
•
t :| M (p) |≥ 1, else
3 ECAPNSim
Active rules development is an activity that needs to be performed carefully. Nowadays,
there are few systems [2] which perform the analysis and debug the ECA rule base.
Most of commercial ADBs [7], [8] provide a syntax to ECA rule definition, however
static analysis of ECA rules cannot be performed inside these systems and the ECA rule
definition is only in a text way.
ECAPNSim (ECA Petri Nets Simulator) was developed under MAC OS X Server
in Java. Taking advantage of Java portability, ECAPNSim can be executed in different
operating systems. As an engine of ECA rules, ECAPNSim can be connected with any
relational database systems such as Postgres, MS Access, Oracle, and Visual Fox Pro.
3.1 ECAPNSim architecture
ECAPNSim architecture consists of two building blocks: ECAPNSim Kernel and ECAP-
NSim tools environment. ECAPNSim Kernel provides active functionality to passive
database. it consists of CCPN Rule Manager, CCPN rule base, Composite Event De-
tector, and Rule Execution Component. ECAPNSim tools environment has a set of tools
used by the ECA rule developers. Tools environment is composed by ECA rule editor,
analyzer of no-termination problem, converter of ECA rules to CCPN, CCPN visual-
izer/editor and explanation components, termination analyzer and runtime tools.
ECAPNSim offers two modalities. In Simulation mode, an user can simulate the
behavior of the ECA rule base modeled by depositing tokens into the CCPN manually.
And, in Real mode, the CCPN is executed by state modification of the connecting DBS.
98

Fig.1. ECAPNSim graphical interface.
3.2 ECAPNSim Design
ECAPNSim offers a graphical and visual interface to represent ECA rule bases by
CCPN model. Like any PN editor, ECAPNSim simulates the behavior of ECA rules
by executing the CCPN model. Meanwhile simulation is running, problems like no-
termination and confluence can be observed obviously in the CCPN, hence ECA rule
developer can modify the rule base to improve it.
The core of ECAPNSim is CCPN models. ECAPNSim contains a module to gener-
ate a CCPN structure from an ECA rule base definition written in a text file automati-
cally. Or a CCPN model can be edited directly from a ECAPNSim user.
ECAPNSim supports CCPN design and edition from an ECA rule base, which can
be moved to another position in the visualization panel. Moreover, because of there
are large ECA rule bases, ECAPNSim will generate large CCPN structures, so it has
zoom buttons to either increase or decrease the CCPN size. Simulation speed can be
controlled through a slide. Finally, the graphical interface has tools and icons to edit a
CCPN, simulate a CCPN behavior, and CCPN file management. (figure 1).
4 Example
In order to illustrate the use of ECAPNSim, a small ECA rule base is presented. The
rules are concerning to a small DB with three tables EMP, BONUS, and SALES. Table
EMP whose attributes are the employee’s id, the employee’s name, the employee’s rank,
and the employee’s salary. Table BONUS whose attributes are the employee’s id for the
corresponding employee, and the bonus amount. Table SALES whose attributes are the
employee’s id, the month when the sales were made, and the number of products that
the employee sold. Table definition is as follows: EMP(emp
id, name, rank, salary),
BONUS(emp
id, amount), SALES(emp id, month, number);
There are three events that will be monitored by the ECA rule base, a) when the
attribute BONUS.amount is modified e0 : update BONUS amount); b) when the at-
tribute EMP.rank is modified (e1 : update
EMP rank), too; and c) when the sales of
a month are registered into the DB and a new record is added in table SALES (e2 :
99

Fig.2. CCPN generated by ECAPNSim.
insert SALES). Four ECA rules are used to illustrate this example, which are related to
DB structure described above. ECA rules description is presented in the next rows:
Rule 1: When an employee’s bonus is increased by more than 100, then the em-
ployee’s rank is increased by 1. Rule 2: When an employee’s rank is updated, then
increased by 1, then the employee’s bonus is increased by 10 times the new rank. Rule
3: When an employee posts sales greater than 50 and its rank lower than 15, his bonus
is decreased by 100. Rule 4: When an employee’s rank reaches 15, increases the em-
ployee’s salary by 10%.
In figure 2 can be observed that event e
0
is depicted by place p
0
, event e
1
is depicted
by place p
1
, and event e
2
is depicted by place p
2
. Places p
3
, p
4
∈ T
copy
because event
e
1
, depicted by p
1
, is used in the firing of two rules (rules 2 and 4). Place p
5
is only
output place, i.e. p
5
represents the action part of rule 4. Transitions t
0
, t
1
, t
2
, t
3
∈ T
rule
store rules 1,2,3, and 4, respectively; and transition t
4
∈ T
copy
is used to replicate
information of event e
1
in order to trigger transitions t
1
and t
3
.
In order to start the CCPN simulation in ECAPNSim, a console application is started
too. Console application establishes a communication process with ECAPNSim to send
the SQL instructions typed in it from an user. After that, ECAPNSim takes the messages
sent by console and convert them in tokens, that are placed in the corresponding CCPN
places. In this case, the simulation is started by typing in the console the instruction:
update bonus set amount = 150 where emp
id = 1. When ECAPNSim receives the
instruction, a token is generated and it is placed in p
1
. Because of p
1
is the input place
of t
1
, then t
1
is activated and fired, t
1
takes token information to evaluate the condition
stored in it, and if it is true according to the random evaluation, then t
1
sends a token
to its output place p
2
. If all transitions t ∈ T
rule
are evaluated to true, then the possible
DB states reachable from an event occurrence can be predicted.
In real mode, the evaluation of condition stored in transitions t ∈ T
rule
is verified
against the DB state. If the instruction typed in console is executed in real mode, then
token generator will put a new token in place p
0
with information about the update
100

command. Condition part in t
0
is evaluated to true, because amount is equal to 150,
and the condition is if amount > 100. Then, token generator creates a new token
according to action part of rule 1 stored in t
0
, where the employee’s rank is increased
by one, i.e. the value for employee’s rank now is 4. New token is sent to the output
place p
1
, which sends the token to t
4
. t
4
replicates the token and it sends one token to
p
3
and the other one to p
4
. Transitions t
1
and t
3
are activated, but they are not triggered
because both conditional parts of these transitions are evaluated to false against the
token information.
5 Conclusion and Future Work
In this research work, an ECA rule base simulator ECAPNSim is developed. ECAP-
NSim uses a Conditional Colored Petri Net (CCPN) as a model to depict ECA rules.
Unlike other ECA rule simulators or DB management systems, ECAPNSim is indepen-
dent on the actual database system, i.e., a developed ECA rule base can work with any
database system the communication bridge ODBC-JDBC between ECAPNSim and its
DBMS. Through the CCPN model, ECAPNSim can simulate an ECA rule base behav-
ior and analyze its static properties. Furthermore, the portability of the programming
language Java makes it possible to connect ECAPNSim with many DBMS in different
operating systems, such as Postgresql in both Linux and Macintosh platforms, and MS
Access, MS Visual Fox Pro, and Oracle, in MS Windows platform, etc.
For future, ECAPNSim will be enhanced with distribution functions such as defin-
ing interarrival times among primitive events.
References
1. Zimmer, D.: ”Rule Termination Analysis based on a Rule Meta Model”. In: Cadlab Report 2,
Cadlab, Bahnhofstr. 32, 33102 Paderborn, Germany, April 1995
2. Schlesinger, M. and L
¨
orincze, G. : ”Rule modeling and simulation in ALFRED”. 3rd Interna-
tional workshop on Rules in Database (RIDS’97), Sk
¨
ovde, Sweden, June, pp. 83-99, 1997
3. Gatziu S., Dittrich K.R., “Detecting Composite Events in Active Database Systems Using
Petri Nets”, Proc. 4th International Workshop on Research Issues in Data Engineering: Active
Database Systems, Houston, Texas, February 1994
4. Li X., Medina Mar
´
ın J., and Chapa S. V., ”A Structural Model of ECA Rules in Active
Database”, Mexican International Conference on Artificial Intelligence 02, M
´
erida, Yucatan,
M
´
exico, April 22-26, 2002
5. Li X., Medina Mar
´
ın J., ”Composite Event Specification in Active Database Systems: A Petri
Net Approach”, IEEE International Conference on System, Man, and Cybernetics, The Hague,
The Netherlands, Oct, 2004.
6. Jensen K., ”An Introduction to the Theoretical Aspects of Colored Petri Nets”. Lecture Notes
in Computer Science: A Decade of Concurrency, vol. 803, edited by J. W. de Bakker, W.-P. de
Roever, G. Rozenberg , Springer-Verlag, 1994, pp. 230-272
7. Hanson E.N., ”The Design and Implementaci
´
on of the Ariel Active Database Rule System”,
IEEE Transactions on Knowledge and Data Engineering, Vol. 8, No. 1, february 1996.
8. Widom J., ”The Starburst Active Database Rule System”, IEEE Transactions on Knowledge
and Data Engineering, Vol. 8, No. 4, August 1996.
101