本研究主要是利用Rule-Based System 對現有存在的WS 進行調適莘取，以提高WS 所提供的價值給使用該系統的使用者。在實際規劃執行流程中，對於如何描述規劃、領域知識、推論出流程中所需任務等問題必須解決，以及在完成取得相關Service 後，調適初步規劃、組合任務成為完整的規劃、描述服務流程等相關議題也須進行處理。我們的研究中是使用AI 技術對planning 進行描述定義、及採用Partial-order planning 演算法進行規劃出流程，利用JESS Inference Engine 機制來實作出Rule-Based 的推論機制。其領域知識則是利用知識工程中的Ontology 技術來描述知識，並使用W3C WebOnt 組識所開發的Web Ontology Language(OWL)來實作出知識庫，以提供在User 進行Planning 動作，有足夠的知識背景可以達到使用者對WS 進行系統流程規劃的目的。而在利用Rule-Based 推論機制及OWL的Ontology 技術成功的將與User 目的必須要執行的Service 任務取出，及進行Planning 完成系統中要執行的規劃流程。在最後的結果，採用Workflow pattern技術作為服務流程結構設計的依據，以確保流程可以正確完整執行。所以系統規劃的過程是一種專業知識的分析與經驗的學習。我們整合 Rule-based system和workflow pattern 的技術，用以規劃網路服務流程。本研究將以規劃整合式旅遊系統為案例，探討該研究方法的可行性，且實作出對於並評估研究成果的效益，則本研究對於網路服務調適取出進行規則式推論，確實在服務流程規劃上有極大的便利。 The integration of heterogonous and distributed systems with Web services(WS) has become the popular solutions in the enterprise for the deployment of electronic business processes. Since the initial use of the WS composition in value-added and composite services building, much effort has been put into further increasing its efficiency and selectivity, while decreasing the amount of building cost. Unfortunately, such development efforts have shown diminishing return in recent years. This has prompted many groups to seek AI planning technologies for the improvement of WS composition. First, we explored the use of ontology for the production of specific domain knowledge. Then, we use state computing and reformed Partial-Order-Planning (POP) algorithm for beforehand planning the corresponding service flow according to user’s request. For the effective control of execution order, we combine the workflow patterns technology. And for the confirmation that the completed plan will adapt to the user’s situation and be supported with the existing services, we combine the mechanism of dynamic service selection. Finally, we build an adaptive WS planning system. From the promising experiments, we find that the dynamic service selection can ensure the validation of composite services and the reformed POP can increase the capability of composite services and decrease the time cost of service composition. Because the characteristics of our domain knowledge creation is domain-independent, the capability of our WS planning can be extended and applied to various application domains by the including of corresponding domain knowledge base.
