| In recent years, the unconventional emergencies were highly valued both here and abroad. Because of their properties like great harmfulness, broad impact, chain reaction and complex developments, they could hardly be handled with general emergency measures. Nowadays the method of case-based reasoning(cbr) was broadly applied in the field of emergency management, using the successful experience s to deal with the current situation. Case representation and retrieval on emergency event chain were key technologies in cbr, and would be mainly researched in this paper.Considering the properties of chain reaction and evolution, a shareable and extensible domain knowledge model on emergency event was founded, which described the general characteristics and evolution relations of emergency events. Then, a meta model based on domain model was founded to describe specific types of events. So the event chain model was given by the instances of meta mode ls and the explicit method of determining evolution relations at last.After the event chain model being given, the similarity algorithms about event properties set, evolution relations and event chain's structure were well defined from three aspects—the difference of event properties, relations and architecture. The comprehensive similarity between event chains was determined by the three similarity algorithms. In according to the event chain similarity, the case with higher event chain similarity would be more likely retrieved out and possessed more valuable information.At last, an emergency case retrieval prototype system was introduced in this paper. The main functions and working methods of the system and the retrieval processes and algorithms were explained in detail. The appropriate modifications to the similarity algorithms were especially discussed. Then, the feasibility of similarity algorithms and the practicability of emergency case retrieval prototype system were well proved by several retrieval instances. |
PDF Full Text Request