Jianguo Lu

Up                                  

Web Service Wrapping, Searching, and Integration

Web service is a new model of distributed computing that is being widely accepted within the software industry. According to Gartner Group, through 2H02, 75 percent of enterprises with greater than $100 million in revenue will interface periodically with web services (0.8 probability). Our goal is to support the migration of existing software systems to web service-based architectures. In particular, we are going to study web service wrapping, discovery, and composing. 

A web service is a self-contained, self-describing, modular, loosely coupled and reusable application that can be published, located, and invoked across the web.  The cornerstone of web service technology is XML. By using meaningful XML tags, the web service is self-describing, and hence easy to be located and reused. Since web service uses XML and HTTP protocols, the service can be accessed across the web.

Conceptually, web service is a new model of distributed computing. In this model, software components will be published, requested and even brokered over the Internet.  Technically, web services are defined in terms of a stack of emerging standards for service description (e.g., WSDL), invocation (e.g., SOAP), publication/discovery (e.g., UDDI), and composition (e.g. WSFL).

Web service technology is motivated by two drawbacks in current software development practice. One is that the plethora of services provided on the web is meant for humans to use, not for applications to access and integrate. The other drawback is that existing distributed component models such as CORBA, DCOM are based on protocols other than HTTP and XML, which means that they are not easy to access over the Internet, or go through the firewalls. Web service technology is meant to combine the best of the two approaches while avoiding the drawbacks.

I am interested in the following issues:

1)     Design web service definition, publication, and composition languages.  Although there are various emerging standards for all those languages, there are many issues undecided and improvements wanted.

2)     Wrapping software components into web services. Most web services come from existing software components instead of being written from scratch. We will investigate the wrapping of web services from various sources, such as HTML web sites, interface definitions of distributed software components such as CORBA or DCOM, and EJB remote objects.

3)     Composing web services. A major advantage of web services is that they can be composed dynamically over the web, relying only on the HTTP protocol. We will study service composition using XML schema inferences and semi-structured data integration techniques.

4)     Discovering web services. We propose to define a similarity metric for web services in terms of the structure of web service signatures and the semantic distance between tag names. The signature matching of web services will depend on XML schema matching, and the definition of semantic distance will use ontologies and other knowledge representation techniques.  

5)     Brokering web services.  The ultimate goal of introducing web services is to promote a new distributed computing model where web services can dynamically collaborate and interoperate to produce new functions. This is accomplished by web service brokers which accept complex requests, locate relevant sub-services from the web, and use those services together to address the initial request.