PhDOOS Logo

Supported by AITO.

These pages are currently maintained by Alex Potanin

 

Web Content Change Management using Agent-Oriented Approach

Ezzeddin HATTAB
( Jordanian )

national technical University of Athens
Stavraetoy 18, Zographou 15772, Athens - Greece
tel: +30 1 7488051
fax: +30 1 7722499
ezz@softlab.ece.ntua.gr

Keywords:

analysis/design formal calculus agents web content change management

Abstract:

Nowadays, organizations drive their businesses via the web that provides new opportunities and competitive threats. They are always in rush to deploy their updated content with high quality in a competitive time. Most of the current web sites have problems with contents that are corrupted, outdated, inconsistent and unauthorized. The key issue of such problems is lack of change management that able to manage the updated content. However, as a step toward the second-generation web systems, some of commercial tools have been developed for web content such as Rational Suite Content Studio, MS SourceSafe, Harvest, PVCS and others. Unfortunately, such tools lack sound theoretical foundation that causes ad hoc solutions. Furthermore, they applied traditional software configuration management that provides static rather than dynamic solutions that should meet the changeable nature of the web. New issues like content expiry, link validations, content state verifications, users and search engines notifications are not automated in such systems. Contrary, the team development that includes site manager, editors, developers and business users are asked to verify such conditions manually that ma! y ! ! ad hoc to their own experience and knowledge of the site. Content change management might be a conflicting term that used in three different trends: configuration management before deployment, managing the content through the browser, and managing the content as a successor state after change actions (our term). To the best of our knowledge, no systems have been developed that able to reason about the change virtually according to a specified pre and post conditions. For example, deploying a new page into application server will put the content into a new state that needs automatic verification that helps the development team to provide consistent, authorized and accurate content. In this project, we will introduce theoretical and practical aspects to build agents that able to reason about new states of the web content. These agents assumed to be autonomous and have a true plug-and-play character to other agent-based web systems. These agents able to detect, validate, store and notify others (human or software agents) about changes. We follow a methodology for agent-oriented analysis and design that describes the problem from an abstract to detailed views. In analysis, we define formal notation that defines positive and negative effects that could be occurred in the change action. We represent dynamically changing content web in which all such changes are the result of defined actions. The web site is conceived as being in some state s, and this state can change in consequence of some agent (human or software) performing an action. If a is some such action, then the successor state to s resulting from the performance of action a denoted by do(a,s). In general, actions may be parameterized in the form of do(a1, do(a2,do(a3,...)...))). Normally, actions will have pre-conditions and post-conditions; sufficient conditions which current state must satisfy before and after the action can be performed in this state. For example, it is illegal to move a page from its location without updating its tag of other pages that point to it. Also, other agents who are interested in the site should be notified as a post-condition of the ! ch! ! ange action. In design, we map formal specifications into a low level abstraction using object-oriented techniques, namely UML. We will identify agents for detection, validation, and notification roles. Also identifying their communication language (ACL) between them and the external world agents. In implementation, we use hybrid technologies (as the nature of web system). Mainly, the development environment is Enterprise Java Beans (EJB) by Sun Microsystems used to implement the core of the specifications and action automations. For content side, we apply three-layered technologies eXtensible Markup Language (XML), Resource Description Framework (RDF) and Ontology Inference layer (OIL) technologies for content instantiations and verification.

The PhD work started: Oct. 1, 1999

HTML3
JAVA

 

Last modified on Mon Aug 15 14:59:24 2005