Designing under the influence of speech acts: a strategy for composing enterprise integration solutions

Designing enterprise-wide integration solutions remains a difficult task. Enterprise Integration Patterns (EIP) [1] provide possible design solutions that may be used to compose enterprise-wide integration solutions. Because of the multitude of platforms on which legacy systems are implemented, these composed solutions must ensure platform-independent implementation, e.g. with web services. A promising mechanism that allows this path is conversation models [2] that may be used to implement interactions among web services that represent different legacy systems. For this translation to occur, though, EIPs must be converted into a representation that is amenable to a conversation models. For example, consider the pattern publish/ subscribe channel [1]. This pattern will need to be defined in terms of speech acts [3] (e.g. informative). A precondition to this informative message exchange would be an agreement on part of publishers to publish such information. Further, the roles played may include: Publisher as Initiator and Sender, and Subscribers as Receiver. A knowledge base that captures the EIPs would, thus, need to capture all, the pre-conditions, the post-conditions, the roles, and the speech acts that comprise the integration pattern. We are developing a research prototype, IDAssist (Integration Designer Assistant), which would include such a knowledge base, and present the designers with the functionality to decide appropriate EIPs for their integration task. This research prototype is being developed using Java™ with an interface that allows drag and drop capability using Java Swing. This interface would allow designers to develop business process diagrams using BPMN [4] and guide them in identifying appropriate pattern for each connections in the business processes. It contains knowledge base that stores integration patterns in an XML specification, called Enterprise Integration Pattern XML (eipXML). Outputs from the tool can be used to design enterprise integration solutions based on EIPs and network of speech acts, converted into appropriate conversation policy specifications [2] that govern interactions among web services.