Emergency management engineering provides the logical and systematic process for reacting to the incidents and determining the treatment opinions. Presently many researches dedicated to the application of case based reasoning (CBR) in emergency engineering. However, usually the ideal cases in emergency domain are not available and hard to obtain. The distance between raw information and cases lies in the middle of information processing and CBR application, making it difficult to design practical CBR systems. Alternatively, this paper proposed an approach of generating imperfect cases from raw information with information evaluation and ranking. First, the derivation of information was assessed. Moreover, to implement that strategy, the technique of natural language process is employed. The gathered information was labeled with keywords, and the keywords gained the weight according to the word frequency and the ontology. Finally, the case was represented by a set keywords with their weights. For many emergency domains, this strategy partly solved the above problem of insufficient cases, and supported the decision making to some extent.
Recently with the increasing occurrence of disasters or instances, emergency management engineering, which provides the logical and systematic process for reacting to the incidents and determining the treatment opinions, becomes necessary and high on the research agendas. Unlike the management engineering in other domain, then emergency management engineering is under the circumstances of inadequate information, changing conditions and
high urgency, thus the methodology of models or rules cannot satisfy the requirement of emergency management. CBR is a methodology to solve new problems with previous experience. It contains four phases, or four “res”, as known as to retrieve the most similar cases, to reuse them, to revise the proposed solution, and to retain the current cases. Since CBR was put out by Schank in 1977, it has achieved great success in amount of fields, such as medical science, information science, industrial context, etc. The heterogeneity of the application domains demonstrates the flexibility and capability of CBR to handle issues which would be too difficult to manage with rules or models [1].
Consequently, a lot of efforts have been devoted to apply CBR (Case-based Reasoning) in emergency engineering.
However, this time CBR does not achieve the expected performance.
The main difficult of CBR in emergency engineering is the organization of disaster or incident cases, which is an essential element in applications of CBR. Usually the design of CBR system assumes that the cases are already
available, and the all of the involved fields are where experiential cases can be easily obtained and collected.
Nevertheless, the domain of emergency engineering is not equivalent in that it concerns amounts of information,such as the description of incident, the involved agencies, the emergency response, the results and evaluation. The information cannot compose a ‘case’ only when they were engineered to be ordered.
This paper focuses on the initial step of CBR, which means the “C”, in other words, the cases. In that he distance
of raw information to the cases is still a vacancy of applying CBR in emergency engineering. The organization of
the paper is as follows. In section 2 we review the current work related with application of CBR and information
processing in emergency engineering. In section 3 we discuss the difficulty of obtaining disaster cases. In section 4
we describe an architecture for information processing to transform information into cases, and demonstrate the
feasibility of the approach with some examples. In section 5 we briefly outline the potential function of the
engineered information. And in section 6 we summarize the ideas discussed throughout the paper.
In this paper we reviewed the current application of CBR in emergency management engineering, and discuss the features of disaster cases. Then we proposed a strategy of information processing to fill out the gap betwe‘case’ and raw information. The raw information was transformed into vector space, within which the vector nor figured out in a new way. With calculating the entropy of words, we selected the keywords and by which the carepresented. For the domain of emergency management engineering without available structured cases, the stratakes use of raw information and keeps the history “as they are”, and is appropriate for the domains where unstructured cases are available.
