IJCSNS International Journal of Computer Science and Network Security, VOL.6 No.5A, May 2006 120 Manuscript received May 5, 2006. Manuscript revised May 25 , 2006. Customized Content Delivery through XML Message Brokering R. Gururaj, and P. Sreenivasa Kumar Indian Institute of Technology Madras, Chennai, India Summary XML is being accepted as a standard format for representation and exchange of web data. XML message brokers play a key role as message exchange points for messages sent between producers and consumers. An XML message broker can perform filtering, transformation, and routing of received messages. In certain applications, XML message brokers may need to perform advanced customization (value-addition) where modifications to the structure and content of the original message are done as per the preferences of individual clients. Customizing the content of messages is desirable and significant in the context of personalized content delivery, data and application integration, and co- operation among disparate web services. At present, XML message brokers support user profile matching and limited customization only. In this paper, we propose a system named VAXBro, a value-adding XML message broker that addresses the data processing needs of value-addition process in an XML message broker. In this work, we also discuss the proposed customization service specification language, XML update approach and process optimization techniques. Key words: Message customization, XML, message broker, application integration. Introduction Information Dissemination involves distributing data produced by data sources to a set of interested data consumers in a distributed environment. A message broker (MB) plays a key role as central exchange point for messages sent between message sources and consumers. In this article we deal with MBs that adopt publish- subscribe (PS) information dissemination model. The PS model is a specialization of information dissemination protocol with push-based, aperiodic data-delivery mechanism. In this paradigm, submitted user preferences are called profiles. The pub-sub system accepts profiles from the end-user, and collects new information/data from different sources and matches received information against profiles and updates the user with relevant information. As Extensible Markup Language (XML) has emerged as a standard for representation and exchange of data on the web, we assume that the future dissemination systems support XML data exchange. A message broker that handles XML data is called an XML message broker. The important functions of an XML message broker are: (a) filtering incoming messages against large number of user queries to find if the message matches the user requirements, (b) transformation that restructures the message according to the user requirements, and (c) routing which involves transmitting the message to the user. The above mentioned transformation activity can customize the message under distribution, to facilitate data and application integration, and personalized content delivery. This kind of customization is different from web- personalization, which refers to the action that changes the layout and content of the web page according to the user preferences. But the personalization in XML message brokers deals with customization of data contained in XML format, which is meant for processing; not just viewing as in the case of HTML. In the recent past, many XML filtering systems have been proposed [1][2][8][9]. All these systems accept user interests (longstanding profiles) in the form of XPath [14] expressions and perform filtering on arrival of a message, to find the set of user queries that match the incoming message. A recent XML message broker proposed in [3] accepts user queries in XQuery [14] format and provides slightly advanced customization functionality where the result of a user query is delivered to the user with customized tags. In customized content delivery scenario, it is appropriate to have XML message brokers that support more advanced transformations that satisfy the needs of end user. Next, in the context of application integration, it is natural to see applications that are developed independently in a distributed environment. In such cases, applications may be dealing with XML data that have structural disparity. When these applications are integrated, one application may send data to the other application for the purpose of further processing. Because of the structural mismatch, the data received can't be used directly. If the data is transformed to suit the needs of receiving application, then application integration becomes more straightforward and effective.