Research On Goal-Driven Context-aware Flexible Web Service Composition

With the development of new information and network techniques, the connection between emergencies and these techniques become more complex and unstable. The frequency and range of this influence is rising greatly. The biggest challenge from Agriculture Emergency Management Information (AMEI) is how to collaborate with corresponding departments to carry out emergency response timely, accurately and fast in order to meet the immediate needs for emergencies. The main idea of Service Composition (SC) focuses on resolving complicated problems through effectively combining the different functions of the service resources. Therefore, the research and development of Composite Service meets the requirements of AMEI by providing technical support. However, the emerging SC methods, which pre-define complete SC plans, cannot meet the need of SC flexibility, such as users’ ambiguous demands and the changing environment during the service execution. Therefore, in order to improve the efficiency and adaptability of SC, the flexibility issues concerning users, resources and environments should be fully considered.Based on the immediate requirements of SC flexibility in different stages of life cycle and the in-depth research of the SC description, automatic generation and dynamic adjustment, we proposed three algorithms and a system to prove these algorithms. This paper consists of four sections:Given that SC correctness and completeness are the basis for the flexibility, we created a new description algorithm based on GC-OWL-S, which utilizes the trinity view (target, Context and service) to extend and improve OWL-S. This new method not only contains goals and contexts of SC, but also fully illustrates the features, such as realm, function and non-function.According to users’ ambiguous demands and the changing environment during the service execution, we designed dynamic programming algorithms and knOWLedge model combined with HTN method.Thus, the service generation has been changed into dynamic programming. We adopted a two-stage policy to generate SC plans automatically, precisely and flexibly by decomposing users’ demands and matching the contextual information. According to possible exceptions during the SC execution, we proposed a dynamic context-aware adjustment algorithm to re-schedule and re-plan SC. This algorithm can guarantee the SC execution.Based on all the algorithms above, we created a multi-layer emergency decision system, including description of the emergency service resources, generation of emergency action plan and the adjustment during the action execution. This decision system proves effectiveness and practicality of all these algorithms.