MANAGING COMPLEX INFORMATION IN REACTIVE
APPLICATIONS USING AN ACTIVE TEMPORAL XML
DATABASE APPROACH
Essam Mansour, Kudakwashe Dube and Bing Wu
School of Computing, Dublin Institute of Technology, Kevin Street, Dublin 8, Ireland
Keywords: Information management, XML, ECA rule paradigm, active database, and temporal database.
Abstract: Tasks, such as those in patient care practice, require constant monitoring of a dynamic context and
environment based on best practices in the form of predefined experience-based information or knowledge.
These applications could take the form of reactive applications. The problem of incorporating best practices
into routines used in such tasks requires advanced approaches and methods for comprehensively managing
complex information. This paper presents a generic and unified framework for Complex Information
Management (CIM) in domains where best practices need to be incorporated into day-to-day work. The
CIM framework is supported by a high level declarative language, called AIM. The approach adopted here
uses the combined application of the event-condition-action (ECA) rule paradigm, a temporal mechanism,
advanced DBMS features and XML technologies. Furthermore, the paper also presents the conceptual
architecture for the complex information management system, AIMS. The main contribution of CIM
framework and approach lays in managing the complex information by introducing multiple management
planes under a unified framework.
1 INTRODUCTION
Reactive applications have the ability to process
events of interest to domain entities. They can
respond to changing situations by issuing alerts,
reminders, requests, and/or observations to users.
The information and knowledge used in guiding the
reactive aspect of applications usually exist in
generic forms as best practice, experience, policies
and operational rules that guide decisions and
actions. This information and knowledge require
formalisation to be incorporated into reactive
applications. Later, they need to be customized and
instantiated to suit the condition or situation in the
domain. The result of the customization and
instantiation process is what we refer to, in this
work, as complex information. This will be made up
of situation-specific, dynamic information and
knowledge that represent the best practice in the
domain. It further forms the basis of the reactive
behaviour that monitors occurrences of relevant
situations in the domain. Examples of this type of
complex information include: the medical patient
plan in the intensive care applications; the customer
order in securities trading; and the customer banking
alerts scenario. The medical patient plan monitors
specific clinical data items and responds to the
changes of these data items with respect to specific
patient according to customized reactive behaviour
based on the clinical guidelines. The security order
deals with particular stock items, to them a set of
constrains and conditions are applied. These
constrains are subject to modification according to
the customer’s experience as response to the stock
items changes, until the security order is executed or
cancelled.
A major challenge in reactive applications is to
provide a unified framework in which the complex
information can be managed as one body in such a
way that the formalisation, specification, execution,
manipulation and query dimensions are explicitly
provided as first class elements under one roof
within the framework.
In the area of active database, managing reactive
aspects of applications is addressed at the level of
rules and triggers, such as in (Umeshwar et al. 1988;
Widom and Ceri 1996; Paton and Diaz 1999;
Bonifati et al. 2002), not at the level of a unified
520
Mansour E., Dube K. and Wu B. (2007).
MANAGING COMPLEX INFORMATION IN REACTIVE APPLICATIONS USING AN ACTIVE TEMPORAL XML DATABASE APPROACH.
In Proceedings of the Ninth International Conference on Enterprise Information Systems - DISI, pages 520-523
DOI: 10.5220/0002385005200523
Copyright
c
SciTePress
body of complex information. Consider, as an
example, the users of securities trading and intensive
care applications demand computerized support for
managing dynamic orders and medical patient plans,
respectively, instead of individual rules and triggers.
Existing support for reactive applications suffer
from lack of support for queries and manipulation.
Furthermore, problems of unexpected interactions
among the individual rules become more likely with
the growing of the rule base (Umeshwar et al. 1988).
This paper presents on-going research (Dube
and Wu 2001; Wu and Dube 2001; Wu and Dube
2001; Dube et al. 2002; Dube et al. 2002; Dube
2004; Dube et al. 2005; Mansour et al. 2006). The
first stage of the research was aimed at providing a
generic Framework with a declarative language
PLAN (Wu and Dube 2001) for managing clinical
reactive applications based on Clinical Practice
Guidelines (CPGs). Also a prototype system TOPS
(Wu and Dube 2001; Dube 2004) was developed to
implement the PLAN framework and language. This
second stage focuses on investigating into a formal
complex information management model and
approach that: 1) is based on XML, 2) takes into
account temporal issues, 3) continue to exploit the
event-driven approach adopted in the previous stage;
and 3) allows advanced queries, replays of past
history and information mining.
2 FRAMEWORK AND
APPRAOCH
This section presents the framework for complex
information management and the adopted approach
to computerising the complex information.
2.1 The CIM Framework
The Complex Information Management (CIM)
framework supports the management of the complex
information in reactive applications. As illustrated in
Figure 1, the CIM framework contains three planes:
specification plane; instantiation and customization
plane and; management plane with the DBMS, XML
technologies and inter-plane communication as the
integrating factors. The DBMS provides support for
the ECA rule paradigm and the temporal
mechanism.
Com plex inform ation
Generation and
Installation
Instantiation and
Customization Plane
Instantiation and
Customization Plane
Specification PlaneSpecification Plane
Complex
Inform ation Kernel
Management Plane
Dissemination
DBMS and XML
Technologies
General
Specification
Query
Distribution
Manipulation
Execution
Figure 1: the Complex Information Management (CIM)
framework.
The complex information management
process is fitted into the three planes of the
framework, as follows:
1) Specification plane: Best practice in domains
for reactive applications is formalised and
translated into formal specifications and held
permanently in the database where they can be
managed easily.
2) Instantiation and customization plane: The
stored generic specifications are used to create
the Complex Information (CI).
3) Management plane: The installed CI is
managed. The plane provides support to
execute, manipulate, query, and disseminate the
CI. The distributed management for the CI
supports the mobility of the domain entities as
well as remote access to the information. An
engine based on the active database is utilized to
execute the CI. The CI is also subject to the
same manipulation operations, as the domain
information. The manipulation operations are
included in the behaviour of CI, or issued by the
domain users. The manipulation operations
facilitate the propagation of the changes from
the generic specification to the CI. The CI also
is subject to the same queries, as the domain
information, plus special query support, such as
replay CI. The replay functionality provides
ability to review the evolution of the CI.
An XML-based language, called AIM, is
developed to support the functionalities of the three
planes. AIM language consists of a model for the
complex information and two sub-languages, AIM-
SL and AIM-QL. AIM-SL and AIM-QL support the
functionalities of the specification plane and the
management plane, respectively. The model of CI
MANAGING COMPLEX INFORMATION IN REACTIVE APPLICATIONS USING AN ACTIVE TEMPORAL XML
DATABASE APPROACH
521
provides support to the instantiation and
customization plane. The approach adopted in
developing AIM language is presented in the next
subsection.
2.2 Active Temporal XML Approach
The Active Temporal XML approach utilizes the
combined application of the Event-Condition-Action
(ECA) rule paradigm, a temporal mechanism,
advanced DBMS features and XML technologies. A
map of the domains involved in this approach is
shown in Figure 2. Based on this approach, AIM
language is developed such that:
1) It is an XML based language and enjoys the
general benefits of XML.
2) It has an active XML-based specification
component language, called AIM-SL, which
utilizes the ECA rule paradigm to formalise and
specify the application best practice. The design
of AIM-SL is based on modularised XML
Schemas. This modularization provides
flexibility in modifying or enriching the AIM-
SL language to suit several application domains.
3) The complex information model is developed as
an active temporal XML document, which
includes the history of the CI and the rules that
represent the behaviour part of CI.
4) It has a high level XQuery-based component
language, AIM-QL, that provides support to:
a) Manipulate both the AIM-SL
speciation and the CI documents;
b) Query and replay the CI;
c) Provides temporal extensions to the
XQuery language to support the replay
functionality.
XML
ECA Rule
Paradigm
Tem poral
Database
Active
XML
Temporal
Trigger
Tem poral
XML
Active Tem poral
XML Database
active
Database
and
Figure 2: The active temporal XML database approach.
3 CONCEPTUAL
ARCHITECTURE
This section presents the conceptual architecture of
the AIM System (AIMS), whose development is
currently on-going. AIMS utilises the AIM language
to manage the complex information according to the
CIM framework.
XML
Schemas
XML Repository
- Domain Information
- AIM-SL Specification
- CI Document
CI Manager
- Specification
- Instantiation
- Execution
- Manipulation
- Query and Replay
Rule Manager
- Rule Execution
- Rule Manipulation
Communication Mana
g
- Dissemination method
- Distribution method
DBMS Trigger
Mechanism
Registration and
Manipulation
Modifications
Information Manager
- Validation method
- Temporal storage method
- Temporal query method
Information
AIM Language
Result
Figure 3: Conceptual architecture of AIMS.
A conceptual architecture of AIMS is presented in
Figure 3. The Complex Information (CI) manager
provides the high level management for the complex
information. CI manager provides support to specify
the application best practice, generate, execute,
manipulate, and query CI documents. The rule
manager provides for the execution of CI rules, their
manipulation and extension of the active mechanism
to support temporal rules. The information manager
extends an XML database system to provide
temporal support and utilizes an XML DBMS to
validate and store the AIM-SL specification and the
CI documents. The communication manager
supports the remote access and distributed
management to the AIM-SL specification and CI
documents. The AIM language is implemented in
the CI manager. AIMS is being implemented by
using DB2, java, and XML technologies, such as
XQuery and Web services.
AIMS avoids the unexpected interactions that
most likely appear with the growing of the rule base,
because:
1) The rules are modularised and joined to a
specific domain entity instead of specific
relation, such as table in the relational database,
and
ICEIS 2007 - International Conference on Enterprise Information Systems
522
2) The rule manager is able to remove a set of rules
according to its objective or scope.
4 CONCLUSION
This paper has presented the Complex Information
Management (CIM) framework and approach, which
consists of three planes for specifying, customizing
and managing the complex information. The AIM
language that is developed to support the
functionalities of the CIM framework planes has
been briefly outlined. The framework also includes a
complex information model that enables easy
performance of information management functions.
The paper has also presented a conceptual
architecture for the AIMS system, which utilizes the
AIM language in managing the complex information
by following the CIM framework. The development
of the prototype system, AIMS, is currently in
progress. The CIM framework and approach are
unique in:
1) Computerizing the complex information as one
distinct entity that is easy to be specified,
executed, manipulated, queried and
disseminated using one language, AIM.
2) Allowing the complex information to be subject
to the same manipulation operations and
queries, as the domain information, plus special
operations and query functionality.
3) Utilizing the generally available highly
optimized and easily maintained technologies,
such as XML technologies and ECA rule
paradigm as incorporated in the modern DBMS,
to provide a tool that assists domain experts and
users in managing the complex information.
REFERENCES
Bonifati, Angela, Daniele Braga, Alessandro Campi, et al.
(2002). Active XQuery. Proceedings of the 19th
International Conference on Data Engineering ICDE,
San Jose (California).
Dube, Kudakwashe (2004). A Generic Approach to
Supporting the Management of Computerised Clinical
Guidelines and Protocols, Dublin Institute of
Technology (DIT).
Dube, Kudakwashe, Essam Mansour and Bing Wu (2005).
Supporting Collaboration and Information Sharing in
Computer-Based Clinical Guideline Management.
18th IEEE Symposium on Computer-Based Medical
Systems (CBMS 2005), Dublin, Ireland, IEEE
Computer Society.
Dube, Kudakwashe, Bing Wu and Jane Grimson (2002).
Using ECA Rules in Database Systems to Support
Clinical Protocols. 13th International Conference on
Database and Expert Systems (DEXA 2002).
Dube, Kudakwashe, Bing Wu and Jane Grimson (2002).
Framework and Architecture for the Management of
ECA Rule-Based Clinical Protocols. 15th IEEE
Symposium on Computer-Based Medical Systems
(CBMS 2002), Maribor, Slovenia, IEEE Computer
Society.
Dube, Kudakwashe and Bing Wu (2001). "Supporting
Clinical Laboratory Test-Ordering Protocol
Specification, Execution and Management: an Event-
Condition-Action Rule and Database Approach."
Healthcare Informatics Journal 7(1): 20-28.
Mansour, Essam, Bing Wu, Kudakwashe Dube, et al.
(2006). An Event-Driven Approach to Computerizing
Clinical Guidelines Using XML. In the First
International Workshop on Event-driven Architecture,
Processing and Systems (EDA-PS’06), In conjunction
with ICWS 2006 / SCC 2006., Chicago, USA, IEEE
Computer Society.
Paton, Norman W. and Oscar Diaz (1999). "Active
database systems." ACM Comput. Surv. 31(1): 63--
103.
Umeshwar, Dayal, Barbara T. Blaustein, Alejandro P.
Buchmann, et al. (1988). "The HiPAC Project:
Combining Active Databases and Timing
Constraints." SIGMOD Record 17: 51-70.
Widom, Jennifer and Stefano Ceri (1996). Active
Database Systems: Triggers and Rules For Advanced
Database Processing, Morgan Kaufmann.
Wu, Bing and Kudakwashe Dube (2001). PLAN: A
Framework and Specification Language with an
Event-Condition-Action (ECA) Mechanism for
Clinical Test Request Protocols. Hawaii International
Conference on System Sciences (HICSS-34), Los
Alamitos, California, IEEE Computer Society.
Wu, Bing and Kudakwashe Dube (2001). Applying Event-
Condition-Action Mechanism in Healthcare: A
Computerized Clinical Test-Ordering protocol System
(TOPS). CODAS.
MANAGING COMPLEX INFORMATION IN REACTIVE APPLICATIONS USING AN ACTIVE TEMPORAL XML
DATABASE APPROACH
523